Code of paper "A new baseline for edge detection: Make Encoder-Decoder great again"
Download the dataset to any dir and point to the dir in the code
-BSDS500 following the setting of "The Treasure Beneath Multiple Annotations: An Uncertainty-aware Edge Detector"
-NYUDv2 following the setting of "Pixel Difference Networks for Efficient Edge Detection" and random crop to 400*400
-BIPED following the setting of "Dense Extreme Inception Network for Edge Detection"
Down it from https://huggingface.co/sail/dl/resolve/main/caformer/caformer_m36_384_in21ft1k.pth and put it into the dir ./model
python main.py --batch_size 4 --stepsize 3-4 --gpu 1 --savedir 0305-bsds --encoder Dul-M36 --decoder unetp --head default --note 'training on BSDS500' --dataset BSDS --maxepoch 6
Following the previous methods. such as RCF and PiDiNet
The result of BSDS500 can be download here https://drive.google.com/file/d/1PiPklsH7w6zNxdGWW-JpnUsFOdiYLHwG/view?usp=sharing