search

lorenmt / mtan

The implementation of "End-to-End Multi-Task Learning with Attention" [CVPR 2019].
https://shikun.io/projects/multi-task-attention-network
MIT License
673 stars 109 forks source link

Cannot unpickle dict_mean_std due to not mentioning python version #39

Closed janhenriklambrechts closed 3 years ago

janhenriklambrechts commented 3 years ago

Hi there, I have tried multiple versions of python to unpickle the dict_mean_std but I keep getting a KeyError: 10 due to trying to unpickle in a different python version. Would you be so kind to tell me either the python version you used to pickle this object or the actual mean and std values. Thanks in advance!

lorenmt commented 3 years ago

Hello,

I am sorry to hear that you are having this problem. I was using python 3.7 but have confirmed 3.8 and 3.9 also work as well. But don't worry, I have copied the mean_std for your convience:

{'ucf101std': array([0.10898684, 0.10810246, 0.10679939]), 
'omniglotstd': array([0.27046935, 0.27046935, 0.27046935]), 
'caltech256mean': array([0.55177064, 0.53382816, 0.50567107]), 
'vgg-petsstd': array([0.25922821, 0.25420126, 0.26146911]), 
'sketchesmean': array([0.97960958, 0.97960958, 0.97960958]), 
'dtdstd': array([0.25226877, 0.2409176 , 0.2499409 ]), 
'vgg-petsmean': array([0.47812268, 0.44583364, 0.39578528]), 
'svhnmean': [0.4376821, 0.4437697, 0.47280442], 
'daimlerpedclsmean': array([0.48203529, 0.48203529, 0.48203529]), 
'mnistmean': array([0.18228742, 0.18228742, 0.18228742]), 
'planktonstd': array([0.16591533, 0.16591533, 0.16591533]), 
'gtsrbstd': array([0.27560347, 0.26576119, 0.27089863]),
 'planktonmean': array([0.94033073, 0.94033073, 0.94033073]), 
'imagenet12mean': [0.485, 0.456, 0.406], 
'vgg-flowersstd': array([0.28509808, 0.23842338, 0.2639633 ]), 
'gtsrbmean': array([0.33921263, 0.3117836 , 0.32047045]), 
'ucf101mean': array([0.49953101, 0.49880386, 0.49981996]), 
'svhnstd': [0.19803012, 0.20101562, 0.19703614], 
'dtdmean': array([0.52696497, 0.47025164, 0.42396662]), 
'mit-indoorstd': array([0.25213846, 0.24675795, 0.24937516]), 
'aircraftstd': array([0.21070221, 0.20508901, 0.23729657]), 
'cifar100mean': [0.50705882, 0.48666667, 0.44078431], 
'mit-indoormean': array([0.48822609, 0.43138942, 0.37296835]), 
'omniglotmean': array([0.08099839, 0.08099839, 0.08099839]), 
'aircraftmean': array([0.47983041, 0.51074066, 0.53437998]), 
'caltech256std': array([0.31026209, 0.30732538, 0.32094492]), 
'sketchesstd': array([0.06532731, 0.06532731, 0.06532731]),
'daimlerpedclsstd': array([0.23612616, 0.23612616, 0.23612616]), 
'imagenet12std': [0.229, 0.224, 0.225], 
'vgg-flowersmean': array([0.43414682, 0.38309883, 0.29714763]), 
'cifar100std': [0.26745098, 0.25647059, 0.27607843], 
'mniststd': array([0.38076293, 0.38076293, 0.38076293])}

Hope that helps.

janhenriklambrechts commented 3 years ago

Thank you so much! I have no clue why this unpickling error happens but this solves my problem. Thanks again