精度问题 - Githubissues

Res2Net / Res2Net-PretrainedModels

(ImageNet pretrained models) The official pytorch implemention of the TPAMI paper "Res2Net: A New Multi-scale Backbone Architecture"

https://mmcheng.net/res2net/

1.07k stars 215 forks source link

精度问题 #27

Closed RLiBIn closed 4 years ago

RLiBIn commented 4 years ago

训练次数410000次，使用test_net.py评价网络效果： MODEL: META_ARCHITECTURE: "GeneralizedRCNN" WEIGHT: "/root/data/Res2Net-maskrcnn/output/model_0100000.pth" BACKBONE: CONV_BODY: "R2-50-FPN" 和 WEIGHT: "/root/data/Res2Net-maskrcnn/output/model_0410000.pth"的评价结果一模一样，这是不合理的

RLiBIn commented 4 years ago

2020-03-11 08:23:45,199 maskrcnn_benchmark.inference INFO: OrderedDict([('bbox', OrderedDict([('AP', 0.7914337039392441), ('AP50', 0.9361981849504082), ('AP75', 0.8667782752642387), ('APs', 0.7299510615202045), ('APm', 0.9113126064292995), ('APl', 0.8986741976921192)])), ('segm', OrderedDict([('AP', 0.6800589144759959), ('AP50', 0.9237830292417535), ('AP75', 0.7669049069704228), ('APs', 0.584172678507268), ('APm', 0.8424472066046184), ('APl', 0.8499092881290765)]))])