Project Title

A web application to extract face and object annotations from videos using computer vision algorithms for face/object detection/recognition

Prerequisites

On a GNU/Linux Debian Stretch machine install

• python 2.7
• docker
• docker-compose

Installing

Clone git repository on your machine. On a console run:

git clone https://github.com/fedjo/orca.git

Move into directory 'aat' and build docker images

cd aat
./docker/build.sh dev

Then run containers using docker-compose

docker-compose up -d

Running the tests

Built With

• Python 2.7 - The language used
• Django 1.10 - Python Web framework

Contributing

Please read CONTRIBUTING.md for details on our code of conduct, and the process for submitting pull requests to us.

Versioning

We use SemVer for versioning. For the versions available, see the tags on this repository.

License

This project is licensed under the MIT License - see the LICENSE.md file for details

Upload a face database

Upload a zip file containing the faces you want to find on videos.

• URL

  /model

• Method:

  POST

• Data Params

  model=[path of the zipfile]

• Success Response:

  • Code: 200
    Content: { "people" : [ "Obama", "Trump" ] }

• Sample Call:

    $.ajax({
      url: "/model",
      dataType: "json",
      type : "POST",
      data: new BinaryData('/path/to/file')
      success : function(r) {
        console.log(r);
      }
    });

List all face databases

List all face databases that the recognizer has been trained with.

• URL

  /model

• Method:

  GET

• Success Response:

  • Code: 200
    Content: [ "actors.yml", "students.yml"" ]

• Sample Call:

    $.ajax({
      url: "/model",
      dataType: "json",
      type : "GET",
      success : function(r) {
        console.log(r);
      }
    });

Annotation Process

Call the annotation process to generate automatic annotations.

• URL

  /annotate

• Method:

  POST

• Data Params

  { "content": {"path": "/thesis_video/I148276[HD-VI].mp4"}, "cascade": { "name": [1], "scale": "1.3", "neighbors": "5", "minx" : "25", "miny":"25", "framerate": 1 }, "bounding_boxes": "True", "objdetector": {"framerate": 50}, "transcription": {"input_language": "it", "output_language": "en"} }

• Success Response:

  • Code: 200
    Content: { "transcription": { "url: /static/I148276[HD-VI].srt" }, "facedetection": { "457.0": [ { "position": { "xaxis": 982, "yaxis": 181 }, "dimensions": { "width": 398, "height": 398 }, "face": "Obama" "probability": 0.56 }, { "position": { "xaxis": 982, "yaxis": 181 }, "dimensions": { "width": 398, "height": 398 }, "face": "Obama" "probability": 0.56 } ], "601.0": [ { "position": { "xaxis": 982, "yaxis": 181 }, "dimensions": { "width": 398, "height": 398 }, }, { "position": { "xaxis": 982, "yaxis": 181 }, "dimensions": { "width": 398, "height": 398 }, } ], }, "objectdetection": { "457.0": [ { "position": { "xaxis": 982, "yaxis": 181 }, "dimensions": { "width": 398, "height": 398 }, "class": "person" "probability": 0.56 }, { "position": { "xaxis": 982, "yaxis": 181 }, "dimensions": { "width": 398, "height": 398 }, "class": "dog" "probability": 0.56 } ], ""601.0"": [ { "position": { "xaxis": 982, "yaxis": 181 }, "dimensions": { "width": 398, "height": 398 }, }, { "position": { "xaxis": 982, "yaxis": 181 }, "dimensions": { "width": 398, "height": 398 }, } ], } }

• Error Response:

  • Code: 400 BAD REQUEST
    Content: { error : "Bad JSON structure" }

  OR

  • Code: 400 BAD REQUEST
    Content: { error : "Input JSON is not appropriate" }

• Sample Call:

    $.ajax({
      url: "/annotate",
      dataType: "json",
      type : "POST",
      data: jsondata
      success : function(r) {
        console.log(r);
      }
    });