search

ShaoqingRen / faster_rcnn

Faster R-CNN
Other
2.71k stars 1.22k forks source link

VOC 2012 Test Results mAP = 0 #89

Open zacurr opened 8 years ago

zacurr commented 8 years ago

I ran the following code 'experiments/script_faster_rcnn_voc2012_vgg16.m'

loss was correctly decreased during training.

However, in the last stage (below), I got Results = 0

fprintf('\n**\nfinal test\n******\n'); model.stage2_rpn.nms = model.final_test.nms; dataset.roidb_test = Faster_RCNN_Train.do_proposal_test(conf_proposal, model.stage2_rpn, dataset.imdb_test, dataset.roidb_test); opts.final_mAP = Faster_RCNN_Train.do_fast_rcnn_test(conf_fast_rcnn, model.stage2_fast_rcnn, dataset.imdb_test, dataset.roidb_test);

When I learn script_faster_rcnn_demo.m with trained result for several images, I got correct results.

Also I found the same thing in the log files you officially provide.

https://onedrive.live.com/download?resid=36FEC490FBC32F1A!110&authkey=!ACpgYZR2MmfklwI&ithint=file%2czip

In this file, results was all zero. It seems that this happens commonly in voc2012_test with other settings. faster_rcnn_VOC2012_vgg_16layers_top-1_nms0_7_top2000_stage2_fast_rcnn/voc_2012_test/log/test_20150824_070207.txt

what is wrong with voc 2012 test?

I am new to voc dataset. please help me.

mumuhe commented 8 years ago

I have the same problem. Have you resolved it?

mystatic commented 7 years ago

Have you solved this problem? I meet this,too.