How to count class when I detect tumor from endoscopy dataset?

[Md_Sagar_Hossen]

I created a project that classifies and predict tumour from an endoscopy video. I trained with an annotation of the point with the tumour, and I tested with video. it is successfully run with the yolo5v algorithm. But now I want to count how many tumours detect in that video. Can anyone help me to count that?
Thanks in advance.

[Mohamed]

Hi, here’s an object counting script that works with Roboflow’s prediction response format:

• roboflow-computer-vision-utilities/object_counting.py at main · roboflow/roboflow-computer-vision-utilities · GitHub

The counting function portion of the script can be adapted to work with detect.py from your YOLOv5 predictions.

• yolov5/detect.py at master · ultralytics/yolov5 · GitHub (processing of results begins after line 137)

[Mohamed]

In addition, to track specific tumors across video frames, for example, you’ll need to implement object tracking.

We also have a GitHub repository for object tracking to help get you started: GitHub - roboflow/zero-shot-object-tracking: Object tracking implemented with the Roboflow Inference API, DeepSort, and OpenAI CLIP.

[Md_Sagar_Hossen]

Thank you sir