search

durandtibo / wildcat.pytorch

PyTorch implementation of "WILDCAT: Weakly Supervised Learning of Deep ConvNets for Image Classification, Pointwise Localization and Segmentation", CVPR 2017
MIT License
267 stars 61 forks source link

The mAP in voc2007 #9

Open xiaopingzeng opened 5 years ago

xiaopingzeng commented 5 years ago

I only get best=93.395 when I run “python3 -m wildcat.demo_voc2007 ../data/voc --image-size 448 --batch-size 16 --lrp 0.1 --lr 0.01 --epochs 20 --k 0.2 --maps 8 --alpha 0.7”. How to get the same result as your paper（voc2007 best mAP=95）?

gaobb commented 5 years ago

I only get 90.607 ( best=91.243 ) which is lower the reslut in paper (95).

$python3 -m wildcat.demo_voc2007 ../data/voc --image-size 448 --batch-size 16 --lrp 0.1 --lr 0.01 --epochs 20 --k 0.2 --maps 8 --alpha 0.7

Epoch: [19] Loss 0.0131 mAP 99.449 Test: 100%|########################################################################################| 310/310 [00:49<00:00, 7.52it/s] Test: Loss 0.1392 mAP 90.607 save model ../expes/models/voc2007/checkpoint.pth.tar *** best=91.243