This program uses an object detection deep learning model and a re-identification model to find and track the objects in a movie. Then, the program will track the trajectory of the objects and check if the objects cross the defined virtual lines or the objects are inside the defined areas. The program uses Intel(r) Distribution of OpenVINO(tm) toolkit for deep learning inferencing.
To track the objects, the program extracts the feature vectors from the objects which consists of 256 FP values with the re-identification DL model. The program can identify which object is most similar to the objects in the previous frame and record the trajectory. The L2 norm (or the distance) of those feature vectores represents the similarity of the objects. This program uses cosine distance to check the similarity between the objects.
The line crossing check and area intrusion check are done by regular (meaning, non deep learning based) arithmetic algorithm.
The latest version playbacks sound when an event such as crossing a line or entering to an area happens.
You can learn how to use line-crossing and area intrusion function from simple test program (test-line-cross-area-intrusion.py
)
このプログラムはディープラーニングの物体検出モデルと特徴抽出(re-identification)モデルを用いて、ムービーフレーム中のオブジェクトを検出し、追跡するプログラムです。プログラムは見つけたオブジェクトの軌跡を記録し、オブジェクトが定義された仮想ラインを超えたり定義されたエリアに侵入したことを検出します。ディープラーニングの推論にはIntel(r) Distribution of OpenVINO(tm) toolkitを利用しています。
オブジェクトの追跡のために、プログラムは特徴抽出モデルを使用して見つけたオブジェクトから256個の浮動小数点数値からなる特徴ベクトルを抽出します。この特徴ベクトル同士のL2ノルム(=距離)がオブジェクト同士の類似度を表します。このプログラムではノルムの計算にコサイン距離を使用しています。
ラインクロスとエリア侵入の判定には通常の(ディープラーニングベースではない)数学的アルゴリズムを使用しています。
最新版はラインクロスしたりエリア侵入したりなどのイベント発生時にオーディオを再生します。
コードを簡素化したテストプログラムも追加しました。ライン交差やエリア侵入判定関数の使い方などを学ぶことが可能です。(test-line-cross-area-intrusion.py
)
[ Update - 04-Apr-2022 ] OpenVINO API 2.0 support. Program is now compatible with OpenVINO 2022.1 (and not compatible with OpenVINO 2021 and older ones)
The demo expects the following models in the Intermediate Representation (IR) format:
- For person / pedestrian detection and re-identification
pedestrian-detection-adas-0002
person-reidentification-retail-0031
You can download these models from OpenVINO Open Model Zoo.
In the models.lst
is the list of appropriate models for this demo that can be obtained via Model downloader
.
Please see more information about Model downloader
here.
- OpenVINO 2021.3
- If you haven't installed it, go to the OpenVINO web page and follow the Get Started guide to do it.
The demo depends on:
opencv-python
numpy
scipy
munkres
To install all the required Python modules you can use:
(Linux) pip3 install -r requirements.txt
(Win10) pip install -r requirements.txt
Use Model Downloader
to download the required models and convert the downloaded model into OpenVINO IR models with Model Converter
.
(Linux) python3 $INTEL_OPENVINO_DIR/deployment_tools/tools/model_downloader/downloader.py --list models.lst
(Win10) python "%INTEL_OPENVINO_DIR%\deployment_tools\tools\model_downloader\downloader.py" --list models.lst
(Linux) python3 object-detection-and-line-cross.py
(Win10) python object-detection-and-line-cross.py
The application draws the results on the screen.
test-line-cross-area-intrusion.py
is a simple test program for line-crossing and area-intrusion detection. You can move mouse cursor to test how it works.
You can utilize coordinates-checker
to get the coordinates of point of interest to define boundary lines and area polygons. The coordinates-checker
is also included in this project as a submodule. See coordinates-checker GitHub repo for details.
python coordinates-checker\coordinates-checker.py [image.jpg / movie.mp4 / 0 ]
- Windows 11 x64
- Intel(r) Distribution of OpenVINO(tm) toolkit 2022.1
- Python 3.7 x64