fasterrcnn-pytorch-training-pipeline

fasterrcnn-pytorch-training-pipeline copied to clipboard

What do the boxes and squares actually indicate?

Hi. Where is this image from?

Need a bit more context.

I cant share the details of the image, but this is my result after running rcnn on a set of these images.

My assumption is that the boxes are the annotations/ground truth I used and the circles are the predictions.

My model is underperforming very highly if that adds anything.

If you are using my repository, then it does not annotate circles. It only annotates bounding boxes on an image. I am not sure what the circles are. But if these are inference results, then I am sure that the boxes are the inference annotations.

I am using your repo. My annotations are all boxes.

By inference annotations you mean the ground truth or predictions?

By annotation, I mean predictions.

Have you faced a circle in your predictions before by any chance?

No. It's very odd.

Hmm, alright. I will leave this issue open for now then, if someone else happens to stumble across the same scenario.