This is a simple program that detects objects in static images or live webcam video feed. It then highlights the bouding boxes and names of those obejcts in unique colors. For object detection the YOLOv2 model, trained on the COCO dataset was used. The COCO dataset contains 80 different object classes. The names of the object classes are available here.
- Download this repository by clicking the green "Clone or download" button on the top left.
- Wait for the Object-detector-master folder to finish downloading (extract the compressed file, if you downloaded as zip).
- Download the YOLOv2 model weights from this link and put the file in the Object-detector-master/yolo-coco/ directory.
- Open the terminal inside the Object-detector-master folder and enter the following command:
python yolo.py. - Select appropriate options for images or webcam feed.
ENJOY!
- You can test the detector on your own custom images by just adding your images to the Object-detector-master/images/ directory.
- You can also save the object detection video and specify the frame rate for the save file (FPS values of 10-30 have been tested).