To address these problems, we present a new boundary-aware cascade network by introducing two novel components.
Ranked #6 on Action Segmentation on GTEA
In this paper, we break the convention of the same training samples for these two heads in dense detectors and explore a novel supervisory paradigm, termed as Mutual Supervision (MuSu), to respectively and mutually assign training samples for the classification and regression head to ensure this consistency.
Spatial downsampling layers are favored in convolutional neural networks (CNNs) to downscale feature maps for larger receptive fields and less memory consumption.
Ranked #88 on Object Detection on COCO minival