Open ForawardStar opened 3 years ago
Hi, I would like to ask you some question about training details. (1) Does your training dataset downloaded from the link : http://mftp.mmcheng.net/liuyun/rcf/data/bsds_pascal_train_pair.lst and http://mftp.mmcheng.net/liuyun/rcf/data/HED-BSDS.tar.gz, or simply from the link: https://www2.eecs.berkeley.edu/Research/Projects/CS/vision/grouping/resources.html. (2) Dose you alter the hyper-parameters of original PyTorch implementation https://github.com/meteorshowers/RCF-pytorch/. When I used the orignal PyTorch implementation , the final F score (0.755) is pretty lower than they reported (0.808), and I do not know why. Look forward to your reply
Hi, I would like to ask you some question about training details. (1) Does your training dataset downloaded from the link : http://mftp.mmcheng.net/liuyun/rcf/data/bsds_pascal_train_pair.lst and http://mftp.mmcheng.net/liuyun/rcf/data/HED-BSDS.tar.gz, or simply from the link: https://www2.eecs.berkeley.edu/Research/Projects/CS/vision/grouping/resources.html. (2) Dose you alter the hyper-parameters of original PyTorch implementation https://github.com/meteorshowers/RCF-pytorch/. When I used the orignal PyTorch implementation , the final F score (0.755) is pretty lower than they reported (0.808), and I do not know why. Look forward to your reply