search

fidler-lab / polyrnn-pp-pytorch

PyTorch training/tool code for Polygon-RNN++ (CVPR 2018)
702 stars 105 forks source link

'ggnn_epoch5_step14000.pt' have a a bit lower mIOU - 69 #42

Open LetsGoFir opened 4 years ago

LetsGoFir commented 4 years ago

I change the data path, edit the path in ggnn.json, and test it with torch1.2/ptyhon3, got a mIoU 69, which is a bit lower than the result published in the paper - 71.38. And I cannot load the 'ggnn_epoch8', can you give me a json file for it? Thanks for you great work!

LetsGoFir commented 4 years ago 

Number of instances
----------------
car: 4517
person: 3239
rider: 537
motorcycle: 148
bicycle: 1129
truck: 93
bus: 98
train: 23
----------------

IOUs
----------------
car: MEAN: 0.7801381199070497 STD: 0.21180478329576186
person: MEAN: 0.7080233897022424 STD: 0.19231650411717952
rider: MEAN: 0.683552623387134 STD: 0.15404026738050183
motorcycle: MEAN: 0.5923347781394875 STD: 0.23855743029070062
bicycle: MEAN: 0.5936849702304672 STD: 0.22490588463020722
truck: MEAN: 0.7594397312370428 STD: 0.19188913178823006
bus: MEAN: 0.8004330341129232 STD: 0.21623632056871203
train: MEAN: 0.5991739700064996 STD: 0.30193679686664776
ALL MEAN: 0.6895975770903557
----------------

N corrections
----------------
car MEAN: 0.0 STD: 0.0
person MEAN: 0.0 STD: 0.0
rider MEAN: 0.0 STD: 0.0
motorcycle MEAN: 0.0 STD: 0.0
bicycle MEAN: 0.0 STD: 0.0
truck MEAN: 0.0 STD: 0.0
bus MEAN: 0.0 STD: 0.0
train MEAN: 0.0 STD: 0.0
ALL MEAN: 0.0
----------------
LetsGoFir commented 4 years ago

And if I skip multi-component objects


IOUs
----------------
car: MEAN: 0.8120224407143601 STD: 0.17786273781972428
person: MEAN: 0.7364667286577709 STD: 0.1642026361339306
rider: MEAN: 0.6996582492880937 STD: 0.14832665964677744
motorcycle: MEAN: 0.6420486222188827 STD: 0.21690518847619442
bicycle: MEAN: 0.6435299546578587 STD: 0.19451260283938904
truck: MEAN: 0.7717278280049291 STD: 0.2065501260934808
bus: MEAN: 0.8789496184596746 STD: 0.14563499165817012
train: MEAN: 0.7870879152792436 STD: 0.27245314275328497
ALL MEAN: 0.7464364196601017
LetsGoFir commented 4 years ago

Do you skip the multi-component objects when evaluation @amlankar? Or maybe there are some problems with my env..

diehualong commented 2 years ago

评估时是否跳过多组件对象@amlankar? 或者也许我的环境有一些问题.. Hi, how did you find the code?