I am wondering if there is a way with the YOLOv3 to select class-agnostic or class-aware object detection. I have seen in other implementations like YOLOv5 setting multi_nms to false, or setting a value called "agnostic" to True in a function.
I see a function in "postprocess.py" called "single_image_nms" and the function in "model.py" for "get_yolo3_inference_model()" links to "batched_yolo3_postprocess". Which is for inferencing... Is this it? If so how do I go about using it? Lastly, is it needed to use this for training as well, and not just inferencing?
I believe this is also referred to as single-label nms/multi-class nms for those more familiar with this concept.
Good Morning,
I am wondering if there is a way with the YOLOv3 to select class-agnostic or class-aware object detection. I have seen in other implementations like YOLOv5 setting multi_nms to false, or setting a value called "agnostic" to True in a function.
I see a function in "postprocess.py" called "single_image_nms" and the function in "model.py" for "get_yolo3_inference_model()" links to "batched_yolo3_postprocess". Which is for inferencing... Is this it? If so how do I go about using it? Lastly, is it needed to use this for training as well, and not just inferencing?
I believe this is also referred to as single-label nms/multi-class nms for those more familiar with this concept.
Thank you for your time