An object detection model build over the Agora's 1-to-1 video call. This model not only allows you to detect the objects over the call but also extract those objects as images.
- Agora.io developer account (with app ID)
- Python3
Jump right into the code and see how to extract objects over agora's video call interface:
- Create an account on agora.io. From the project manager create a new project and use the given app ID for this video call.
- Open Object Detection folder and install all the python requirements through the terminal command
pip install -r requirements.txt
- Open ObjectDetection.py and paste your app ID over there along with a suitable channel name.
- Download the yolo dataset from here(https://github.com/OlafenwaMoses/ImageAI/releases/download/1.0/resnet50_coco_best_v2.0.1.h5) and paste it to the ObjectDetection directory
- Now open a sample agora video call channel and paste the same app ID and channel name over there and click join
- Run ObjectDetection.py - this will take the frames from the video call and apply object detection using ImageAI.
- All the extracted objects can be seen in the folder named - "test_output.png-objects"