dvlab-research/ECCV22-P3AFormer-Tracking-Objects-as-Pixel-wise-Distributions

Why not YOLO for Object Detection?

duckaivn opened this issue · 1 comments

Why not YOLO for Object Detection?

It's a good question. We start our approach under the transformer setting because transformers are really hot nowadays. However, we believe our proposed tracking paradigm may also generalize to YOLO-like detectors (e.g. by only modifying some heads of YOLO detectors). We are now trying this idea and we will add this to the repo as long as we get some new results.