This small tutorial creates a data pipeline from Apache Kafka over MongoDB into R or Python. It focuses on simplicity and can be seen as a baseline for similar projects.

• docker
• docker-compose
• MongoDB Compass

docker-compose up -d

It starts:

• Zookeeper
• Kafka Broker
• Kafka Producer
  • built docker image executing fat JAR
• Kafka Connect
  • with MongoDB Connector
• MongoDB
• RStudio
• Jupyter Notebook

The Kafka Producer produces fake events of a driving truck into the topic truck-topic in JSON format every two seconds. Verify that data is produced correctly:

docker-compose exec broker bash
kafka-console-consumer --bootstrap-server broker:9092 --topic truck-topic

We use Kafka Connect to transfer the data from Kafka to MongoDB. Verify that the MongoDB Source and Sink Connector is added to Kafka Connect correctly:

curl -s -XGET http://localhost:8083/connector-plugins | jq '.[].class'

Start the connector:

curl -X POST -H "Content-Type: application/json" --data @MongoDBConnector.json http://localhost:8083/connectors | jq

Verify that the connector is up and running:

curl localhost:8083/connectors/TestData/status | jq

Start MongoDB Compass and create a new connection with:

username: user
password: password
authentication database: admin
or
URI: mongodb://user:password@localhost:27017/admin

You should see a database TruckData with a collection truck_1 having data stored.

Open RStudio via:

localhost:8787

The username is user and password password.

Under /home you can run GetData.R. It connects to MongoDB using the package mongolite and requests the data.

Get Jupyter Notebooks’ URL:

docker logs jupyter

Under /work you can run the Jupyter notebook using the distribution PyMongo.

• Confluent Docker-Compose file
• Confluent Docker Configuration Parameters
• Confluent Kafka Connect
• Confluent Hub MongoDB Connector
• MongoDB Connector Configuration Properties
• mongolite Package
• PyMongo Distribution

This extends Angular CLI to improve the i18n extraction and merge workflow. New/removed translations are added/removed from the target translation files and translation states are managed. Additionally, translation files are normalized (whitespace, stable sort) so that diffs are easy to read (and translations in PRs might actually get reviewed 😉 ).

Prerequisites: i18n setup with defined target locales in angular.json – as documented here.

ng add ng-extract-i18n-merge

Run ng update ng-extract-i18n-merge@2 to upgrade from v1 to v2. This migration switches to Angular’s built-in extract-i18n builder and changes some defaults (see breaking changes below).

If you plan to upgrade from v1 straight to v3 you must first upgrade to v2 using the command above and then run ng update ng-extract-i18n-merge again for the v3 update.

Breaking changes:

• Now this plugin uses the default Angular extract-i18n target – so you can simply run ng extract-i18n.
• Default sort is now stableAppendNew (was idAsc).
• Leading/trailing whitespaces are normalized (collapsed to one space) but not completely trimmed.
• The provided npm run script was removed (you can create your own if needed).

Run ng update ng-extract-i18n-merge to update to v3.0.0 using the Angular update mechanism. This release drops support for Angular 19 and older. The defaults for prettyNestedTags and sort changed to false and "stableAlphabetNew" respectively. builderI18n now defaults to @angular/build:extract-i18n instead of @angular-devkit/build-angular:extract-i18n. During ng update existing builder configurations are updated to keep the previous behaviour (except for builderI18n, where the new default is best for most setups).

ng extract-i18n # yes, same as before - this replaces the original builder

In your angular.json the target extract-i18n that can be configured with the following options:

Feedback and PRs always welcome 🙂

Before developing complex changes, I’d recommend opening an issue to discuss whether the indented goals match the scope of this package.

• My pypi VT package was transfered to VirusTotal ownership

Before using the tool you must set your api key in one of this file or you can start without creating it and you will be prompted to provide the data:

• Home Directory:

  • ~.vtapi, ~vtapi.conf

• or current directory where vt script placed

  • .vtapi, vtapi.conf

• ~.vtapi file content:

[vt]
apikey=your-apikey-here
type=public
intelligence=False
#coma separated engine list, can be empty
engines=
timeout=60
# as for weblogin, this only required for rule management
username=
password=

• your type of api access, if private: type=private, if public, you can leave it empty, it will be automatically recognized as public
• if you have access to VT Intelligence, you need set intelligence=True

Dependencies:

• requests
• texttable
• python-dateutil

These can be installed via PIP or a package manager.
Example of installing all dependencies using pip:

pip install -r requirements.txt

• Thanks to @kellewic and @urbanski
• Special thanks to @Seifreed for testing and reporting bugs

Few public API functions taken from Chris Clark script
And finally has been added full public and private API support by Andriy Brukhovetskyy (doomedraven)

Small manual with examples
http://www.doomedraven.com/2013/11/script-virustotal-public-and-private.html

• BEAR IN MIND THIS IS AN OLD EXAMPLE, use -h for current help

vt -h
usage: value [-h] [-fi] [-udb USERDB] [-fs] [-f] [-fr] [-u] [-ur] [-d] [-i]
             [-w] [-s] [-si] [-et] [-rai] [-itu] [-cw] [-dep] [-eo] [-snr]
             [-srct] [-tir] [-wir] [-rbgi] [-rbi] [-agi] [-dbc] [-ac] [-gc]
             [--get-comments-before DATE] [-v] [-j] [--csv] [-rr] [-rj] [-V]
             [-r] [--delete] [--date DATE] [--period PERIOD] [--repeat REPEAT]
             [--notify-url NOTIFY_URL] [--notify-changes-only] [-wh] [-wht]
             [-pdns] [--asn] [-aso] [--country] [--subdomains]
             [--domain-siblings] [-cat] [-alc] [-alk] [-opi] [--drweb-cat]
             [-adi] [-wdi] [-tm] [-wt] [-bd] [-wd] [-du] [--pcaps] [--samples]
             [-dds] [-uds] [-dc] [-uc] [-drs] [-urs] [-pe]
             [-esa SAVE_ATTACHMENT] [-peo] [-bh] [-bn] [-bp] [-bs] [-dl]
             [-nm NAME] [-dt DOWNLOAD_THREADS] [--pcap] [--clusters]
             [--distribution-files] [--distribution-urls] [--before BEFORE]
             [--after AFTER] [--reports] [--limit LIMIT] [--allinfo] [--rules]
             [--list] [--create FILE] [--update FILE] [--retro FILE]
             [--delete_rule DELETE_RULE] [--share]
             [--update_ruleset UPDATE_RULESET] [--disable DISABLE]
             [--enable ENABLE]
             [value [value ...]]

Scan/Search/ReScan/JSON parse

positional arguments:
  value                 Enter the Hash, Path to File(s) or Url(s)

optional arguments:
  -h, --help            show this help message and exit
  -fi, --file-info      Get PE file info, all data extracted offline, for work
                        you need have installed PEUTILS library
  -udb USERDB, --userdb USERDB
                        Path to your userdb file, works with --file-info
                        option only
  -fs, --file-search    File(s) search, this option, don't upload file to
                        VirusTotal, just search by hash, support linux name
                        wildcard, example: /home/user/*malware*, if file was
                        scanned, you will see scan info, for full scan report
                        use verbose mode, and dump if you want save already
                        scanned samples
  -f, --file-scan       File(s) scan, support linux name wildcard, example:
                        /home/user/*malware*, if file was scanned, you will
                        see scan info, for full scan report use verbose mode,
                        and dump if you want save already scanned samples
  -fr, --file-scan-recursive
                        Recursive dir walk, use this instead of --file-scan if
                        you want recursive
  -u, --url-scan        Url scan, support space separated list, Max 4 urls (or
                        25 if you have private api), but you can provide more
                        urls, for example with public api, 5 url - this will
                        do 2 requests first with 4 url and other one with only
                        1, or you can specify file filename with one url per
                        line
  -ur, --url-report     Url(s) report, support space separated list, Max 4 (or
                        25 if you have private api) urls, you can use --url-
                        report --url-scan options for analyzing url(s) if they
                        are not in VT data base, read preview description
                        about more then max limits or file with urls
  -d, --domain-info     Retrieves a report on a given domain (PRIVATE API
                        ONLY! including the information recorded by
                        VirusTotal's Passive DNS infrastructure)
  -i, --ip-info         A valid IPv4 address in dotted quad notation, for the
                        time being only IPv4 addresses are supported.
  -w, --walk            Work with domain-info, will walk through all detected
                        ips and get information, can be provided ip parameters
                        to get only specific information
  -s, --search          A md5/sha1/sha256 hash for which you want to retrieve
                        the most recent report. You may also specify a scan_id
                        (sha256-timestamp as returned by the scan API) to
                        access a specific report. You can also specify a space
                        separated list made up of a combination of hashes and
                        scan_ids Public API up to 4 items/Private API up to 25
                        items, this allows you to perform a batch request with
                        one single call.
  -si, --search-intelligence
                        Search query, help can be found here -
                        https://www.virustotal.com/intelligence/help/
  -et, --email-template
                        Table format template for email
  -ac, --add-comment    The actual review, you can tag it using the "#"
                        twitter-like syntax (e.g. #disinfection #zbot) and
                        reference users using the "@" syntax (e.g.
                        @VirusTotalTeam). supported hashes MD5/SHA1/SHA256
  -gc, --get-comments   Either a md5/sha1/sha256 hash of the file or the URL
                        itself you want to retrieve
  --get-comments-before DATE
                        A datetime token that allows you to iterate over all
                        comments on a specific item whenever it has been
                        commented on more than 25 times. Token format
                        20120725170000 or 2012-07-25 17 00 00 or 2012-07-25
                        17:00:00
  -v, --verbose         Turn on verbosity of VT reports
  -j, --dump            Dumps the full VT report to file (VTDL{md5}.json), if
                        you (re)scan many files/urls, their json data will be
                        dumped to separated files
  --csv                 Dumps the AV's detections to file (VTDL{scan_id}.csv)
  -rr, --return-raw     Return raw json, in case if used as library and want
                        parse in other way
  -rj, --return-json    Return json with parts activated, for example -p for
                        passive dns, etc
  -V, --version         Show version and exit

All information related:
  -rai, --report-all-info
                        If specified and set to one, the call will return
                        additional info, other than the antivirus results, on
                        the file being queried. This additional info includes
                        the output of several tools acting on the file (PDFiD,
                        ExifTool, sigcheck, TrID, etc.), metadata regarding
                        VirusTotal submissions (number of unique sources that
                        have sent the file in the past, first seen date, last
                        seen date, etc.), and the output of in-house
                        technologies such as a behavioural sandbox.
  -itu, --ITW-urls      In the wild urls
  -cw, --compressedview
                        Contains information about extensions, file_types,
                        tags, lowest and highest datetime, num children
                        detected, type, uncompressed_size, vhash, children
  -dep, --detailed-email-parents
                        Contains information about emails, as Subject, sender,
                        receiver(s), full email, and email hash to download it
  -eo, --email-original
                        Will retrieve original email and process it
  -snr, --snort         Get Snort results
  -srct, --suricata     Get Suricata results
  -tir, --traffic-inspection
                        Get Traffic inspection info
  -wir, --wireshark-info
                        Get Wireshark info
  -rbgi, --rombios-generator-info
                        Get RomBios generator info
  -rbi, --rombioscheck-info
                        Get RomBiosCheck info
  -agi, --androidguard-info
                        Get AndroidGuard info
  -dbc, --debcheck-info
                        Get DebCheck info, also include ios IPA

Rescan options:
  -r, --rescan          Allows you to rescan files in VirusTotal's file store
                        without having to resubmit them, thus saving
                        bandwidth, support space separated list, MAX 25
                        hashes, can be local files, hashes will be generated
                        on the fly, support linux wildmask
  --delete              A md5/sha1/sha256 hash for which you want to delete
                        the scheduled scan
  --date DATE           A Date in one of this formats (example: 20120725170000
                        or 2012-07-25 17 00 00 or 2012-07-25 17:00:00) in
                        which the rescan should be performed. If not specified
                        the rescan will be performed immediately.
  --period PERIOD       Period in days in which the file should be rescanned.
                        If this argument is provided the file will be
                        rescanned periodically every period days, if not, the
                        rescan is performed once and not repeated again.
  --repeat REPEAT       Used in conjunction with period to specify the number
                        of times the file should be rescanned. If this
                        argument is provided the file will be rescanned the
                        given amount of times, if not, the file will be
                        rescanned indefinitely.

File scan/Rescan shared options:
  --notify-url NOTIFY_URL
                        An URL where a POST notification should be sent when
                        the scan finishes.
  --notify-changes-only
                        Used in conjunction with --notify-url. Indicates if
                        POST notifications should be sent only if the scan
                        results differ from the previous one.

Domain/IP shared verbose mode options, by default just show resolved IPs/Passive DNS:
  -wh, --whois          Whois data
  -wht, --whois-timestamp
                        Whois timestamp
  -pdns, --resolutions  Passive DNS resolves
  --asn                 ASN number
  -aso, --as-owner      AS details
  --country             Country
  --subdomains          Subdomains
  --domain-siblings     Domain siblings
  -cat, --categories    Categories
  -alc, --alexa-cat     Alexa category
  -alk, --alexa-rank    Alexa rank
  -opi, --opera-info    Opera info
  --drweb-cat           Dr.Web Category
  -adi, --alexa-domain-info
                        Just Domain option: Show Alexa domain info
  -wdi, --wot-domain-info
                        Just Domain option: Show WOT domain info
  -tm, --trendmicro     Just Domain option: Show TrendMicro category info
  -wt, --websense-threatseeker
                        Just Domain option: Show Websense ThreatSeeker
                        category
  -bd, --bitdefender    Just Domain option: Show BitDefender category
  -wd, --webutation-domain
                        Just Domain option: Show Webutation domain info
  -du, --detected-urls  Just Domain option: Show latest detected URLs
  --pcaps               Just Domain option: Show all pcaps hashes
  --samples             Will activate -dds -uds -dc -uc -drs -urs
  -dds, --detected-downloaded-samples
                        Domain/Ip options: Show latest detected files that
                        were downloaded from this ip
  -uds, --undetected-downloaded-samples
                        Domain/Ip options: Show latest undetected files that
                        were downloaded from this domain/ip
  -dc, --detected-communicated
                        Domain/Ip Show latest detected files that communicate
                        with this domain/ip
  -uc, --undetected-communicated
                        Domain/Ip Show latest undetected files that
                        communicate with this domain/ip
  -drs, --detected-referrer-samples
                        Undetected referrer samples
  -urs, --undetected-referrer-samples
                        Undetected referrer samples

Process emails:
  -pe, --parse-email    Parse email, can be string or file
  -esa SAVE_ATTACHMENT, --save-attachment SAVE_ATTACHMENT
                        Save email attachment, path where to store
  -peo, --parse-email-outlook
                        Parse outlook .msg, can be string or file

Behaviour options:
  -bh, --behaviour      The md5/sha1/sha256 hash of the file whose dynamic
                        behavioural report you want to retrieve. VirusTotal
                        runs a distributed setup of Cuckoo sandbox machines
                        that execute the files we receive. Execution is
                        attempted only once, upon first submission to
                        VirusTotal, and only Portable Executables under 10MB
                        in size are ran. The execution of files is a best
                        effort process, hence, there are no guarantees about a
                        report being generated for a given file in our
                        dataset. a file did indeed produce a behavioural
                        report, a summary of it can be obtained by using the
                        file scan lookup call providing the additional HTTP
                        POST parameter allinfo=1. The summary will appear
                        under the behaviour-v1 property of the additional_info
                        field in the JSON report.This API allows you to
                        retrieve the full JSON report of the files execution
                        as outputted by the Cuckoo JSON report encoder.
  -bn, --behavior-network
                        Show network activity
  -bp, --behavior-process
                        Show processes
  -bs, --behavior-summary
                        Show summary

Download options:
  -dl, --download       The md5/sha1/sha256 hash of the file you want to
                        download or txt file with .txt extension, with hashes,
                        or hash and type, one by line, for example: hash,pcap
                        or only hash. Will save with hash as name, can be
                        space separated list of hashes to download
  -nm NAME, --name NAME
                        Name with which file will saved when download it
  -dt DOWNLOAD_THREADS, --download-threads DOWNLOAD_THREADS
                        Number of simultaneous downloaders

Additional options:
  --pcap                The md5/sha1/sha256 hash of the file whose network
                        traffic dump you want to retrieve. Will save as
                        hash.pcap
  --clusters            A specific day for which we want to access the
                        clustering details, example: 2013-09-10
  --distribution-files  Timestamps are just integer numbers where higher
                        values mean more recent files. Both before and after
                        parameters are optional, if they are not provided the
                        oldest files in the queue are returned in timestamp
                        ascending order.
  --distribution-urls   Timestamps are just integer numbers where higher
                        values mean more recent urls. Both before and after
                        parameters are optional, if they are not provided the
                        oldest urls in the queue are returned in timestamp
                        ascending order.

Distribution options:
  --before BEFORE       File/Url option. Retrieve files/urls received before
                        the given timestamp, in timestamp descending order.
  --after AFTER         File/Url option. Retrieve files/urls received after
                        the given timestamp, in timestamp ascending order.
  --reports             Include the files' antivirus results in the response.
                        Possible values are 'true' or 'false' (default value
                        is 'false').
  --limit LIMIT         File/Url option. Retrieve limit file items at most
                        (default: 1000).
  --allinfo             will include the results for each particular URL scan
                        (in exactly the same format as the URL scan retrieving
                        API). If the parameter is not specified, each item
                        returned will only contain the scanned URL and its
                        detection ratio.

Rules management options:
  --rules               Manage VTI hunting rules, REQUIRED for rules management
  --list                List names/ids of Yara rules stored on VT
  --create FILE         Add a Yara rule to VT (File Name used as RuleName
  --update FILE         Update a Yara rule on VT (File Name used as RuleName
                        and must include RuleName
  --retro FILE          Submit Yara rule to VT RetroHunt (File Name used as
                        RuleName and must include RuleName
  --delete_rule DELETE_RULE
                        Delete a Yara rule from VT (By Name)
  --share               Shares rule with user
  --update_ruleset UPDATE_RULESET
                        Ruleset name to update
  --disable DISABLE     Disable a Yara rule from VT (By Name)
  --enable ENABLE       Enable a Yara rule from VT (By Name)

Khoshnus is just a library, but it’s also your tool for bringing life to static text in an artistic way. With its
elegant SVG animations, your text can now be revealed as if written by hand!

Khoshnus lets you:

• Animate SVG text with a beautiful, handwritten effect.
• Choose from unique fonts with customizable animations.
• Control each letter’s timing, stroke, and fill for a more personalized touch.
• Transform plain text into an elegant visual experience.

• Multiple Font Styles: Choose from an array of fonts.
• Smooth SVG Animation: Stroke and fill effects are animated for a lifelike handwriting experience.
• Fine-Grained Control: Adjust timing and style for each letter—make them reveal at your own pace.
• Fully Configurable: Every visual aspect of the text can be customized.

To get started with Khoshnus, add the npm package to your project by running either of the following commands:

npm install khoshnus
--------------------
yarn add khoshnus

Do this if you are using React.

import { FONT_MATRIX, initialize, write } from "khoshnus"
import 'khoshnus/style.css'

const App = () => {
    useEffect(() => {
        const manuscript = new Manuscript();
        manuscript.setup({
            font: FONT_MATRIX["Pinyon Script"].name,
            fontSize: "10px",
        });
        const textId = manuscript.write("Hello Universe, My Name Is Optimus Prime!")
        manuscript.erase(textId, { delayOperation: 15000 })
    }, []);

    return (
        <div>
            <svg id="khoshnus" width="100%" height="500" viewBox="0 0 100 100" xmlns="http://www.w3.org/2000/svg"></svg>
        </div>
    )
}

Add an empty SVG element with the id of khoshnus that Khoshnus will use to animate your text. Feel free to adjust the size of the SVG
based on your needs.

<svg id="khoshnus" width="100%" height="100%" viewBox="0 0 100 100" xmlns="http://www.w3.org/2000/svg"></svg>

Once you have your bare HTML file, import the basic stuff needed from the library:

import { initialize, write, FONT_MATRIX } from "khoshnus";

import "khoshnus/style.css";

Then start using the library:

// Create the Manuscript instance and setup global configuration.
const manuscript = new Manuscript();
manuscript.setup({
    font: FONT_MATRIX["Pinyon Script"].name,
    fontSize: "10px",
});
// Write the letters into the SVG element.
const textId = manuscript.write("Hello Universe, My Name Is Optimus Prime!")

// Erase the text after the desired waiting period - optional, you can leave it forever ;)
manuscript.erase(textId, { delayOperation: 15000 })

Khoshnus offers full control over your animation. Want the text to feel like it’s written slowly or quickly? You decide!
Here’s a glimpse of how you can tweak it:

initialize({
    font: FONT_MATRIX["Pinyon Script"].name, // Only fonts from FONT_MATRIX are available.
    fontSize: "16px",
    start: {
        startStrokeDashoffset: FONT_MATRIX["Pinyon Script"].strokeDashoffset,
        startStroke: "black",
        startStrokeWidth: 0.0000000001,
        startFill: "transparent",
    },
    end: {
        endStrokeDashoffset: 0,
        endStroke: "transparent",
        endStrokeWidth: 0.3,
        endFill: "black",
    },
    durations: {
        strokeDashoffsetDuration: 3500,
        strokeWidthDuration: 2500,
        strokeDuration: 2500,
        fillDuration: 4000,
    },
});

The magic of Khoshnus lies in its ability to provide positioning and delay control. Here’s an example:

write("Your Text Here", {
    writeConfiguration: {
        eachLetterDelay: 250, // Delay of each letter after the previous one.
        delayOperation: 0 // Delay of the write operation - useful when you want wait time between written snippets.
    },
    textConfiguration: {
        x: "50%", // X position of the text.
        y: "50%", // Y position of the text.
        textAnchor: "middle", // Anchor of the text.
        dominantBaseline: "middle", // Baseline of the text - where it should align.
        fontSize: "12px" // Font size - appearance could possibly depend on the parent element.
    }
});

Check out the following snippet:

const manuscript = new Manuscript();
manuscript.setup({
    font: FONT_MATRIX["Pinyon Script"].name,
    fontSize: "10px",
});
const textId0 = manuscript.write("Do not lament my absence,", {
    textElementAttributes: { y: "10%", },
    writeConfiguration: { eachLetterDelay: 100 }
})
const textId1 = manuscript.write("for in my spark,", {
    textElementAttributes: { y: "25%" },
    writeConfiguration: {
        delayOperation: 3000,
        eachLetterDelay: 100
    }
})
const textId2 = manuscript.write("I know that this is not the end,", {
    textElementAttributes: { y: "37.5%" },
    writeConfiguration: {
        delayOperation: 5500,
        eachLetterDelay: 100,
    }
})
const textId3 = manuscript.write("but merely,", {
    textElementAttributes: { y: "55%" },
    writeConfiguration: {
        delayOperation: 9500,
        eachLetterDelay: 100,
    }
})
const textId4 = manuscript.write("a new beginning.", {
    textElementAttributes: { y: "67.5%", },
    writeConfiguration: {
        delayOperation: 11500,
        eachLetterDelay: 100,
    }
})
const textId5 = manuscript.write("- Optimus Prime", {
    textElementAttributes: {
        x: "80%",
        y: "90%",
    }, writeConfiguration: {
        delayOperation: 14000,
        eachLetterDelay: 100,
    }
})
manuscript.erase(textId0, { delayOperation: 20000 });
manuscript.erase(textId1, { delayOperation: 20000 });
manuscript.erase(textId2, { delayOperation: 20000 });
manuscript.erase(textId3, { delayOperation: 20000 });
manuscript.erase(textId4, { delayOperation: 20000 });
manuscript.erase(textId5, { delayOperation: 20000 });

It generates the following piece of art:

It is possible to define multiple SVGs in your component/view as long as they have different ids and you define which Manuscript object references which SVG element. The following React.js snippet shows two SVG elements being referenced, each by its corresponding Manuscript object:

const App = () => {
  useEffect(() => {
    const manuscript1 = new Manuscript({
      svgId: "khoshnus-1",
      font: FONT_MATRIX["Pinyon Script"].name,
      fontSize: "10px",
    });
    const text1 = manuscript1.write("Do not lament my absence,", {
      textElementAttributes: {
        y: "10%",
      }, writeConfiguration: { eachLetterDelay: 100 }
    })
    manuscript1.erase(text1)
    
    const manuscript2 = new Manuscript({
      svgId: "khoshnus-2",
      font: FONT_MATRIX["Pinyon Script"].name,
      fontSize: "10px",
    });
    const text2 = manuscript2.write("For in my spark,", {
      textElementAttributes: {
        y: "10%",
      }, writeConfiguration: { eachLetterDelay: 100 }
    })
    manuscript2.erase(text2)
  }, []);

  return (
    <div>
      <svg id="khoshnus-1" width="100%" height="300" viewBox="0 0 100 100" xmlns="http://www.w3.org/2000/svg"></svg>
      <svg id="khoshnus-2" width="100%" height="300" viewBox="0 0 100 100" xmlns="http://www.w3.org/2000/svg"></svg>
      </div>
  )
}

Here are some of the unique fonts you can play with:

• BlackCherry: Bold strokes with an offset of 80
• Celtic: Celtic-inspired, with a stroke offset of 50
• Eutemia: Classic, elegant font with a 60 offset
• Kingthings: Royal and medieval vibes, stroke offset of 40
• Ruritania: Majestic, with an exaggerated offset of 280
• VTKS: Artistic flair, with an offset of 150
• Parisienne: Soft and feminine, offset of 100
• Sevillana: Spanish-inspired curves, with an offset of 120
• Pinyon Script: Formal and sophisticated, offset of 100

Here are just a few ways you can use Khoshnus:

• Display personalized signature animations for your website.
• Add a dynamic handwritten greeting to your home page.
• Animate text for art projects, digital invitations, or logos.
• Create a storybook-like experience with flowing, hand-drawn text.

This project is licensed under the MIT License, meaning you’re free to use, modify, and distribute it in both personal
and commercial projects.

Paldesk SDK supports iOS 10+ and Swift 5.

Currently only Carthage is supported.
Add github "paldesk/paldesk-ios-sdk" to your Cartfile and run carthage update --platform ios

Drag the frameworks (Paldesk, ReSwift, SocketIO, Starscream, Kingfisher and Alamofire from Carthage/Build) to Linked Frameworks and Libraries section.

Add new Run Script phase in Build Phases and add:

/usr/local/bin/carthage copy-frameworks

Input files:

$(SRCROOT)/Carthage/Build/iOS/Kingfisher.framework
$(SRCROOT)/Carthage/Build/iOS/ReSwift.framework
$(SRCROOT)/Carthage/Build/iOS/SocketIO.framework
$(SRCROOT)/Carthage/Build/iOS/Starscream.framework
$(SRCROOT)/Carthage/Build/iOS/Paldesk.framework

Output files:

$(BUILT_PRODUCTS_DIR)/$(FRAMEWORKS_FOLDER_PATH)/Kingfisher.framework
$(BUILT_PRODUCTS_DIR)/$(FRAMEWORKS_FOLDER_PATH)/ReSwift.framework
$(BUILT_PRODUCTS_DIR)/$(FRAMEWORKS_FOLDER_PATH)/SocketIO.framework
$(BUILT_PRODUCTS_DIR)/$(FRAMEWORKS_FOLDER_PATH)/Starscream.framework
$(BUILT_PRODUCTS_DIR)/$(FRAMEWORKS_FOLDER_PATH)/Paldesk.framework

Add NSPhotoLibraryUsageDescription to your Info.plist. It’s needed for uploading images.

Initialize Paldesk with your Api Key in app delegate:

// Swift
import Paldesk

    func application(_ application: UIApplication, didFinishLaunchingWithOptions launchOptions: [UIApplicationLaunchOptionsKey: Any]?) -> Bool {
        // Override point for customization after application launch.
        
        PaldeskSDK.initialize(apiKey: "Your Api Key")

        return true
    }

// Objective-C
#import <Paldesk/Paldesk.h>

- (BOOL)application:(UIApplication *)application didFinishLaunchingWithOptions:(NSDictionary *)launchOptions {
    // Override point for customization after application launch.

    [PaldeskSDK initializeWithApiKey:@"Your Api Key"];

    return YES;
}

If you wish to enable logging, just add “loggingEnabled” parameter to initialize method:

// Swift
    PaldeskSDK.initialize(apiKey: "Your Api Key", loggingEnabled: true)

// Objective-C
    [PaldeskSDK initializeWithApiKey:@"Your Api Key" loggingEnabled:true];

To start conversation from your app, use:

// Swift
    PaldeskSDK.startConversation(viewController: self)

// Objective-C
    [PaldeskSDK startConversationWithViewController: self];

Depending on your Client authentification type settings from Webapp -> Administration -> iOS SDK (https://www.paldesk.com/), you can choose three options:

• No data required – client will appear as “visitor” – you are free to start conversation without setting information about your client
• You provide information about client through code – you can create client when your client becomes available to you with following method:

// Swift
    let params = [
        "email":"clientemail@mail.com",
        "externalId": "123456",
        "firstName": "First Name",
        "lastName": "Last Name"
    ]
    PaldeskSDK.createClient(params: params)

// Objective-C
    [PaldeskSDK createClientWithParams:@{
                                         @"email": @"test@fksdaodfkspo.com",
                                         @"externalId": @"testfasfasdf",
                                         @"firstName": @"First Name",
                                         @"lastName": @"Last Name"
                                         }];

Email and ExternalId fields are mandatory and other fields are optional. ExternalId should be something unique from your side, preferably your client’s id (or even email).
If client information is not provided, your client will appear as “visitor”. You can explicitly create “visitor” by calling:

// Swift
    PaldeskSDK.createAnonymousClient()

// Objective-C
    [PaldeskSDK createAnonymousClient];

• Client provides his information through form – your user will have to provide his information in registration form which will be shown on first start (prior to starting conversation).

To clear (logout) current client, use:

// Swift
    PaldeskSDK.clear()

// Objective-C
    [PaldeskSDK clear];

Important: If you call create client method multiple times, it will matter only first time. If you wish to create different client, please call clear method first and then you are able to create client again.

CHEMMISOL is a performant chemical equilibrium solver built in the context of the CAMMISOL project.

Widely inspired by the VMinteq software its main purpose is to dynamically solve the equilibrium state of chemical systems.

• Feature comparison with VMinteq
• Compilation and installation
  • Compile
  • Test
  • Install
• Specification of chemical systems
  • Reactions
  • Components and species
• Usage
  • Basic usage
  • Setting and fixing pH and quantities of other chemical components

VMinteq is considered as a state of the art reference in the context of chemical speciation computation. However, it is an old fashioned closed source Windows GUI application that cannot be integrated in other models.

The purpose of CHEMMISOL is to provide equivalent features in a flexible, performant, open source, cross platform and embeddable way.

The following table represents a non-exhaustive list of features supported by VMinteq (except for the “Solver algorithm”), and how they are currently supported by CHEMMISOL.

Even if CHEMMISOL already allows to solve equilibriums in complex chemical systems, a lot of features are still missing, and will probably never be included, depending on our own needs.

[1] K. Meintjes and A. P. Morgan, “A methodology for solving chemical equilibrium systems,” Applied Mathematics and Computation, vol. 22, no. 4, pp. 333–361, Jun. 1987, doi:10.1016/0096-3003(87)90076-2.

The library can be compiled as a regular CMAKE project.

git clone https://github.com/u2worm/chemmisol-cpp.git
cd chemmisol-cpp
cmake -DCMAKE_BUILD_TYPE=Release -B build -S .
cmake --build build

To run tests:

cd build/tests
ctest

or:

cd build/tests
./chemmisol-tests

The library can be installed on the current system with:

cd build
cmake --install .

Depending on your OS and installation, you might need to run sudo cmake --install .

The definition of chemical systems in CHEMMISOL is based on the concept of components, chemical species and reactions.

Basically, to define a chemical system, it is required to:

• Define a set of reactions.
• Define components and their total quantities.

Components are the building blocks of reactions: the purpose of reactions is to specify how chemical components can combine to form complex chemical species.

Chemical components basically represent canonical species that cannot be divided. A reaction must be specified as a set of reagents, associated to stoichiometric coefficients. By convention:

• Reactants are specified with positive coefficients.
• Products are specified with negative coefficients.

All reactants and products must correspond to components of the chemical system, except one reagent, that represents the produced species of the reaction.

Reactions:

1. H2 <-> H + H
2. 2 H2O <-> 4 H + O2

Components:

1. H
2. H2O (solvent)

The specification of this system is correct. The produced species of reaction 1 is then H2, and the produced species of reaction 2 is O2.

Reactions:

1. H2 <-> H + H
2. 2 H2O <-> 2 H2 + O2

Components:

1. H2
2. O2

The specification of this system is correct. The produced species of reaction 1 is then H, and the produced species of reaction 2 is H2O.

Reactions:

1. H2 <-> H + H
2. 2 H2O <-> 4 H + O2

Components:

1. H2
2. H2O (solvent)

The specification of this system is wrong, since two chemical species that are not components are identified in reaction 2: H and O2. In consequence, the produced species of reaction 2 cannot be properly identified.

A chemical species is a concrete physical entity that lives in the chemical system and can interact with other species. Each chemical species is associated to a quantity (typically in mol), a concentration (typically in mol/l) and an activity (without unit, used in the computation of reaction quotients).

As seen in the reaction example above, chemical species can be defined in two ways:

• Implicitly, as produced species of reactions.
• Explicitly, by declaring components.

Indeed, each chemical component is automatically associated to a chemical species with the same name. However, the nature of a component is different from the nature of chemical species, since it represents the total quantity of species constituted from this component.

For example, let’s consider the following system:

Reactions:

1. H2O <-> OH- + H+
2. PO4 + 3H+ <-> H3PO4

Components:

1. H+
2. PO4
3. H2O (solvent)

Produced species are then OH- and H3PO4.

The total quantity N of the PO4 and H+ components are then defined as follows:

• N(PO4) = n(PO4) + n(H3PO4)
• N(H+) = n(H+) - n(OH-) + 3 * n(H3PO4)

where n denotes the quantity of each chemical species. Notice that n(PO4) and n(H+) denote the quantities of PO4 and H+ species, that are likely not equal to total quantities of PO4 and H+ components. See the ChemicalComponent documentation for more details about how the total quantities are computed.

In practice, the user input of the model represents the total quantity of each component. The actual quantity of each species is then determined by the solved equilibrium state of the chemical system.

The CHEMMISOL API aims to be as simple as possible, in order to facilitate its integration within other projects and models. Here is a basic usage example:

#include "chemmisol.h"

using namespace chemmisol;

int main(int, char *[])
{
    ChemicalSystem chemical_system;
    // Defines the reaction H2O <-> OH- + H+ (log K = -13.997)
    chemical_system.addReaction("OH-", -13.997, {
            {"OH-", -1},
            {"H+", -1},
            {"H2O", 1}
            });
    // Defines the reaction Na+ + Cl- <-> NaCl (log K = -0.3)
    chemical_system.addReaction("NaCl", -0.3, {
            {"NaCl", -1},
            {"Na+", AQUEOUS, 1}, // AQUEOUS specified here only for
                                 // demonstration purpose, since it should be
                                 // the default phase
            {"Cl-", AQUEOUS, 1}
            });
    // Defines the reaction H2O + Na+ <-> NaOH + H+ (log K = -13.897)
    chemical_system.addReaction("NaOH", -13.897, {
            {"NaOH", -1},
            {"H+", -1},
            {"Na+", 1},
            {"H2O", 1}
            });

    // Defines the Na+ component and sets its total concentration to 0.1 mol/l
    chemical_system.addComponent("Na+", 0.1*mol/l);
    // Defines the Cl- component and sets its total concentration to 0.1 mol/l
    chemical_system.addComponent("Cl-", 0.1*mol/l);
    // Defines the H2O component as a solvent
    chemical_system.addSolvent("H2O");

    // Automatically adds the H+ component and fixes the pH to 7
    chemical_system.fixPH(7);

    // Solves the equilibrium state
    chemical_system.solveEquilibrium();
    return 0;
}

This basic example is available in the examples directory and can be run with:

cd build
./examples/basic_chemical_system/basic_chemical_system_example

Expected output:

[chemmisol-core] I Solving chemical equilibrium.
[chemmisol-core] I Init activities:
[chemmisol-core] I  (C) Na+: 0.1
[chemmisol-core] I  (C) Cl-: 0.1
[chemmisol-core] I  (C) H2O: 1
[chemmisol-core] I  (C) H+: 1e-07
[chemmisol-core] I      OH-: 0
[chemmisol-core] I      NaCl: 0
[chemmisol-core] I      NaOH: 0
[chemmisol-core] I Solved activities:
[chemmisol-core] I  (C) Na+: 0.0954352
[chemmisol-core] I  (C) Cl-: 0.0954352
[chemmisol-core] I  (C) H2O: 1
[chemmisol-core] I  (C) H+: 1e-07
[chemmisol-core] I      OH-: 1.00693e-07
[chemmisol-core] I      NaCl: 0.00456476
[chemmisol-core] I      NaOH: 1.20979e-08

The pH of the chemical system can be fixed with the fixPH() method.

However, it is also possible to initialize the pH of the system using the initPH() method. In this case, the pH is set as the total quantity of the H+ component (or any other user defined component eventually specified as argument by the user). The pH is then dynamically determined depending on the solved chemical equilibrium.

As fixing the pH is equivalent to fixing the concentration of H+, the concentration of any component can be fixed in CHEMMISOL, using the fixComponent() method.

The total quantity of a component can also be initially specified using the addComponent() method.

One of the main advantage of CHEMMISOL is that the total quantities of component, fixed or not, including the pH, can be specified dynamically, calling again one of the fixPH(), fixComponent(), initPH(), setTotalQuantity() or setTotalConcentration() methods.

#include "chemmisol.h"

using namespace chemmisol;

int main(int, char *[])
{
    ChemicalSystem chemical_system;
    // Defines the reaction H2O <-> OH- + H+ (log K = -13.997)
    chemical_system.addReaction("OH-", -13.997, {
            {"OH-", -1},
            {"H+", -1},
            {"H2O", 1}
            });
    // Defines the reaction Na+ + Cl- <-> NaCl (log K = -0.3)
    chemical_system.addReaction("NaCl", -0.3, {
            {"NaCl", -1},
            {"Na+", AQUEOUS, 1}, // AQUEOUS specified here only for
                                 // demonstration purpose, since it should be
                                 // the default phase
            {"Cl-", AQUEOUS, 1}
            });
    // Defines the reaction H2O + Na+ <-> NaOH + H+ (log K = -13.897)
    chemical_system.addReaction("NaOH", -13.897, {
            {"NaOH", -1},
            {"H+", -1},
            {"Na+", 1},
            {"H2O", 1}
            });

    // Defines the Na+ component and sets its total concentration to 0.1 mol/l
    chemical_system.addComponent("Na+", 0.1*mol/l);
    // Defines the Cl- component and sets its total concentration to 0.1 mol/l
    chemical_system.addComponent("Cl-", 0.1*mol/l);
    // Defines the H2O component as a solvent
    chemical_system.addSolvent("H2O");

    // Automatically adds the H+ component and fixes the pH to 7
    chemical_system.fixPH(7);

    // Sets up a basic CSV output
    std::ofstream csv_file("output.csv");
    csv_file << "i, pH, Na+, Cl-, NaCl, NaOH"
        << std::endl;

    for(std::size_t i = 0; i < 10; i++) {
        // Makes the ph vary from 7 to 8,
        chemical_system.fixPH(7 + i*0.1);
        // Reduces the total quantity of Na+ from 0.1 mol/l to 0.09 mol/l
        chemical_system.setTotalConcentration(
                chemical_system.getComponent("Na+"),
                0.1*mol/l - i*0.01*mol/l);

        // Solves the new equilibrium state, from the previous equilibrium
        chemical_system.solveEquilibrium();

        // Concentration output (notice getSpecies() is used, not getComponent())
        csv_file << i << "," <<
            chemical_system.getPH() << "," <<
            chemical_system.getSpecies("Na+").concentration() << "," <<
            chemical_system.getSpecies("Cl-").concentration() << "," <<
            chemical_system.getSpecies("NaCl").concentration() << "," <<
            chemical_system.getSpecies("NaOH").concentration() << std::endl;
    }
}

This example is available in the examples directory and can be run with:

cd build
./examples/dynamic_equilibrium/dynamic_equilibrium_example

The output data should look as follows:

Reactants output	Products output (log scale)

This project includes helper functions for working with dictionaries in Monkey C.

The utils include:

• compareDictionaries() that checks if two dictionaries are equal (by comparing their keys and values)
• mergeDictionaries() that adds keys/values of two dictionaries

As any other barrel, in order to use this one, developers will need to add it to the current project.
This can be done through the
Visual Studio Code’s Monkey C Extention
(“Monkey C: Configure Monkey Barrel” option) or by manually editing a project’s Jungle and manifest file.

See “How to Include Barrels” or
“Monkey Barrel Management”
for details.

Once it has been added to a project, the compiler starts “to see” the barrel.
There is no need for importing it, unless you want to use alias for the library.

You can obtain the compiled .barrel file from project’s releases page.

This function adds or overwrites values from one dictionary into another dictionary.

The syntax is:

var result = DictionaryUtilsBarrel.mergeDictionaries(dict1, dict2);

In this case, values from dict2 will be copied to dict1. For convenience, dict1 is returned as function result.

Please, note that dict1 object will change.

Let’s take a look at some basic examples, to illustrate the funcionality:

    var dict1 = { "a" => 1, "b" => 2, "c" => 3 };
    var dict2 = { "d" => 4, "e" => 5, "f" => 6 };

    var result = DictionaryUtilsBarrel.mergeDictionaries(dict1, dict2);

The code above produces the following result:

{"a"=>1, "b"=>2, "c"=>3, "d"=>4, "e"=>5, "f"=>6}

More advanced example

    var dict1 = { "a" => 1, "b" => { "k1" => "v1", "k2" => "v2" } };
    var dict2 = { "b" => { "k2" => "v3" } };

    var result = DictionaryUtilsBarrel.mergeDictionaries(dict1, dict2);

will result with:

 { "a" => 1, "b" => { "k1" => "v1", "k2" => "v3" } }

This function compares two dictionaries and returns true if they are equal and false otherwise.
The function test value equality. The objects here are considered equal, when they have the same keys and corresponding values.

ATTOW, Connect IQ SDK (6.4.2) doesn’t provide reliable way for comparing dictionaries by value. For example, { "a1" => 1, "a2" => 2 } and { "a2" => 2, "a1" => 1 } are two different dictionaries (when you compare them with equals()). Please note, that the keys and corresponding values are the same.

The syntax is:

var result = DictionaryUtilsBarrel.compareDictionaries(dict1, dict2);

A few examples:

    var dict1 = { "a" => 1, "b" => 2, "c" => 3 };
    var dict2 = { "b" => 2, "a" => 1, "c" => 3 };

    var result = DictionaryUtilsBarrel.compareDictionaries(dict1, dict2);

The code above returns true.

    var dict1 = { "a" => 1, "b" => 2, "c" => 3 };
    var dict2 = { "b" => 2, "a" => 1 };

    var result = DictionaryUtilsBarrel.compareDictionaries(dict1, dict2);

returns false as dict2 doesn’t contain element c .

    var dict1 = { "a" => 1, "b" => { "k1" => "v1", "k2" => "v2" } };
    var dict2 = { "a" => 1, "b" => { "k1" => "v1", "k2" => "different" } };

    var result = DictionaryUtilsBarrel.compareDictionaries(dict1, dict2);

will result with false (k2 has different value).

You can build and export the barrel using the IDE (i.e. “Monkey C: Export Project” or “Monkey C: Build Current Project” commands of Monkey C Extention for VS Code) or with the following command:

export VERSION="0.1.0"
barrelbuild -o bin/dictionary-utils-$version.barrel -f monkey.jungl

See: “Exporting a Monkey Barrel” for details.

The project has no dependencies.

If you want to contribute to this project, please fork the repository and create a pull request, or open an issue for discussion.

This project is licensed under the Apache License 2.0. This means you can freely use, modify, and distribute it, under the terms of the Apache License. You can read the full text of the license at http://www.apache.org/licenses/LICENSE-2.0.

This is a functional fork of the Text-to-Clip project.

• We ease the installation and provide a docker environment to run this project.

• If this fork save you precious hours of painful installation, please cite our work:

  @article{EscorciaDJGS2018,
  author    = {Victor Escorcia and
               Cuong Duc Dao and
               Mihir Jain and
               Bernard Ghanem and
               Cees Snoek},
  title     = {Guess Where? Actor-Supervision for Spatiotemporal Action Localization},
  journal   = {CoRR},
  volume    = {abs/1804.01824},
  year      = {2018},
  url       = {http://arxiv.org/abs/1804.01824},
  archivePrefix = {arXiv},
  eprint    = {1804.01824}
  }
  

  TODO: update bibtex with Corpus-Moment-Retrieval-work

• If you like the installation procedure, give us a ⭐ in the github banner.

1. Install docker and nvidia-docker.

   Please follow the installation instructions of your machine.
   As long as you can run docker hello-world container and test nvidia-smi with a cuda container, you are ready to go.

   • Let’s test that you are ready by pulling out our docker image

     docker run --runtime=nvidia -ti escorciavkaust/caffe-python-opencv:latest caffe device_query --gpu 0

     You should read the information of your GPU in your terminal.

2. Let’s go over the installation procedure without the headache of compilation errors.

   • Let’s use a snapshot of the code with less headaches

     git clone git@github.com:escorciav/Text-to-Clip_Retrieval.git
     git checkout cp-functional-testing

   • Then, launch a container from the root folder of the project.

     docker run --runtime=nvidia --rm -v /etc/passwd:/etc/passwd -u $(id -u):$(id -g) -v $(pwd):$(pwd) -w $(pwd) -ti escorciavkaust/caffe-python-opencv:latest bash

   • In case, you are not in the working directory. Move to that folder.

     Make sure that you replace the [...] with your filesystem structure when you copy-paste the command below 😅.

     cd [...]/Text-to-Clip_Retrieval

   • Follow the instructions outlined here from step 2 onwards.

That’s all, two simple steps to get yourself up and running 😉.

1. I close the container. Do I need to repeat the installation steps?

   Nope. All the libraries reside inside your root folder not in the image.

2. I close the container. How can I launch it again?

   Go to the root folder and type

   docker run --runtime=nvidia --rm -v /etc/passwd:/etc/passwd -u $(id -u):$(id -g) -v $(pwd):$(pwd) -w $(pwd) -ti escorciavkaust/caffe-python-opencv:latest bash

3. I wanna use pass my own data, and it is in a different folder. How can I access it from the container?

   Let’s assume your data is in your /scratch/awesome-dataset

   docker run --runtime=nvidia --rm -v /etc/passwd:/etc/passwd -u $(id -u):$(id -g) -v $(pwd):$(pwd) -v /scratch/awesome-dataset:/awesome-dataset -w $(pwd) -ti escorciavkaust/caffe-python-opencv:latest bash

   You will find it in /awesome-dataset inside the container.

4. I can’t find the Text-to-Clip_Retrieval folder inside the container.

   Most probably, you were not in the root folder when you launched it.

   Make sure that you replace the [...] with your filesystem structure when you copy-paste the command below.

   cd [...]/Text-to-Clip_Retrieval
   docker run --runtime=nvidia --rm -v /etc/passwd:/etc/passwd -u $(id -u):$(id -g) -v $(pwd):$(pwd) -w $(pwd) -ti escorciavkaust/caffe-python-opencv:latest bash

5. Can you add the Text-to-Clip binaries to the docker image?

   Why not? gimme a ⭐ in the github banner and I will make time for that. The more stars I get, the priority increases.

This is a 3rdparty project, I used git to keep track different changes.

• The default branch devel corresponds to a derivative work related to Corpus Moment Retrieval Project.

  If this branch is useful for you, we would appreciate that you cite our work:

  @article{EscorciaDJGS2018,
  author    = {Victor Escorcia and
               Cuong Duc Dao and
               Mihir Jain and
               Bernard Ghanem and
               Cees Snoek},
  title     = {Guess Where? Actor-Supervision for Spatiotemporal Action Localization},
  journal   = {CoRR},
  volume    = {abs/1804.01824},
  year      = {2018},
  url       = {http://arxiv.org/abs/1804.01824},
  archivePrefix = {arXiv},
  eprint    = {1804.01824}
  }
  

  TODO: update bibtex with Corpus-Moment-Retrieval-work

• The original project is on the master branch.

• A functional version of the Text-to-Clip project that is expected to run without issues is on the cp-functional-testing branch.

Original README 👇

Code released by Huijuan Xu (Boston University).

We address the problem of text-based activity retrieval in video. Given a
sentence describing an activity, our task is to retrieve matching clips
from an untrimmed video. Our model learns a fine-grained similarity metric
for retrieval and uses visual features to modulate the processing of query
sentences at the word level in a recurrent neural network. A multi-task
loss is also employed by adding query re-generation as an auxiliary task.

Our code is released under the MIT License (refer to the LICENSE file for
details).

If you find our paper useful in your research, please consider citing:

@inproceedings{xu2019multilevel,
title={Multilevel Language and Vision Integration for Text-to-Clip Retrieval.},
author={Xu, Huijuan and He, Kun and Plummer, Bryan A. and Sigal, Leonid and Sclaroff,
Stan and Saenko, Kate},
booktitle={AAAI},
year={2019}
}

1. Installation
2. Preparation
3. Train Proposal Network
4. Extract Proposal Features
5. Training
6. Testing

1. Clone the Text-to-Clip_Retrieval repository.

   git clone --recursive git@github.com:VisionLearningGroup/Text-to-Clip_Retrieval.git

2. Build Caffe3d with pycaffe (see: Caffe installation
   instructions).

   Note: Caffe must be built with Python support!

cd ./caffe3d

# If have all of the requirements installed and your Makefile.config in
  place, then simply do:
make -j8 && make pycaffe

3. Build lib folder.

   cd ./lib
   make

1. We convert the orginal data annotation files into json format.

   # train data json file
   caption_gt_train.json
   # test data json file
   caption_gt_test.json

2. Download the videos in Charades
   dataset and extract frames at 25fps.

1. Generate the pickle data for training proposal network model.

   cd ./preprocess
   # generate training data
   python generate_roidb_modified_freq1.py

2. Download C3D classification pretrain model to ./pretrain/ .

3. In root folder, run proposal network training:

   bash ./experiments/train_rpn/script_train.sh

4. We provide one set of trained proposal network model weights.

1. In root folder, extract proposal features for training data and save as
   hdf5 data.

   bash ./experiments/extract_HDF_for_LSTM/script_test.sh

1. In root folder, run:

   bash ./experiments/Text_to_Clip/script_train.sh

1. Generate the pickle data for testing the Text_to_Clip model.

   cd ./preprocess
   # generate test data
   python generate_roidb_modified_freq1_full_retrieval_test.py

2. Download one sample model to ./experiments/Text_to_Clip/snapshot/ .

   One Text_to_Clip model on Charades-STA dataset is provided in:
   caffemodel
   .

   The provided model has Recall@1 (tIoU=0.7) score ~15.6% on the
   test set.

3. In root folder, generate the similarity scores on the test set and save
   as pickle file.

   bash ./experiments/Text_to_Clip/test_fast/script_test.sh

4. Get the evaluation results.

   cd ./experiments/Text_to_Clip/test_fast/evaluation/
   bash bash.sh

Common web and node.js grunt tasks/lint configs/md templates and so on…

• Overview
• Usage
• Installation
• API Documentation
• Contributing
• Release History
• License

This library holds common build tasks, lint configuration and so on, used by my personal node.js and web projects.
This allows for a central configuration for all projects.

In order to use this library, you need to refactor your build processes as defined in some of my projects such as simple-oracledb

In order to use this library, just run the following npm install command:

npm install --save-dev js-project-commons

See full docs at: API Docs

See contributing guide

Developed by Sagie Gur-Ari and licensed under the Apache 2 open source license.