Which YOLOV8 layer to use to check the model's prediction

rigvedrs / YOLO-V8-CAM

Wanna know what your model sees? Here's a package for applying EigenCAM on the new YOLO V8 model

MIT License

130 stars 26 forks source link

Which YOLOV8 layer to use to check the model's prediction #19

Open worstcodergg opened 4 months ago

worstcodergg commented 4 months ago

How do i know which target layer i should look at if i want to find out the closest image output to what the model is predicting whenever it sees an image?

Currently the default target layer for cls is model.model.model[-2], is there a reason for that?