error occurs while training

[berkdenizi]

Epoch: [10/60][400/404] Time 0.173 (0.188) Data 0.000 (0.011) Loss 1.6979 (1.5605) Acc 81.25 (91.09) Lr 0.000300 eta 1:03:19

Evaluating market1501 (source)

Extracting features from query set ...
Done, obtained 3368-by-2048 matrix
Extracting features from gallery set ...
Done, obtained 15913-by-2048 matrix
Speed: 0.0211 sec/batch
Computing distance matrix with metric=euclidean ...
Computing CMC and mAP ...

"ValueError Traceback (most recent call last)
in
4 eval_freq=10,
5 print_freq=10,
----> 6 test_only=False
7 )

~\Desktop\deep-person-reid-master\deep-person-reid-master\torchreid\engine\engine.py in run(self, save_dir, max_epoch, start_epoch, print_freq, fixbase_epoch, open_layers, start_eval, eval_freq, test_only, dist_metric, normalize_feature, visrank, visrank_topk, use_metric_cuhk03, ranks, rerank)
141 save_dir=save_dir,
142 use_metric_cuhk03=use_metric_cuhk03,
--> 143 ranks=ranks
144 )
145 self._save_checkpoint(epoch, rank1, save_dir)

~\Desktop\deep-person-reid-master\deep-person-reid-master\torchreid\engine\engine.py in test(self, epoch, dist_metric, normalize_feature, visrank, visrank_topk, save_dir, use_metric_cuhk03, ranks, rerank)
225 use_metric_cuhk03=use_metric_cuhk03,
226 ranks=ranks,
--> 227 rerank=rerank
228 )
229

~\anaconda3\envs\torchreid\lib\site-packages\torch\autograd\grad_mode.py in decorate_no_grad(*args, **kwargs)
47 def decorate_no_grad(*args, **kwargs):
48 with self:
---> 49 return func(*args, **kwargs)
50 return decorate_no_grad
51

~\Desktop\deep-person-reid-master\deep-person-reid-master\torchreid\engine\engine.py in _evaluate(self, epoch, dataset_name, query_loader, gallery_loader, dist_metric, normalize_feature, visrank, visrank_topk, save_dir, use_metric_cuhk03, ranks, rerank)
300 q_camids,
301 g_camids,
--> 302 use_metric_cuhk03=use_metric_cuhk03
303 )
304

~\Desktop\deep-person-reid-master\deep-person-reid-master\torchreid\metrics\rank.py in evaluate_rank(distmat, q_pids, g_pids, q_camids, g_camids, max_rank, use_metric_cuhk03, use_cython)
199 return evaluate_cy(
200 distmat, q_pids, g_pids, q_camids, g_camids, max_rank,
--> 201 use_metric_cuhk03
202 )
203 else:

~\Desktop\deep-person-reid-master\deep-person-reid-master\torchreid\metrics\rank_cylib\rank_cy.pyx in torchreid.metrics.rank_cylib.rank_cy.evaluate_cy()
22
23 # Main interface
---> 24 cpdef evaluate_cy(distmat, q_pids, g_pids, q_camids, g_camids, max_rank, use_metric_cuhk03=False):
25 distmat = np.asarray(distmat, dtype=np.float32)
26 q_pids = np.asarray(q_pids, dtype=np.int64)

~\Desktop\deep-person-reid-master\deep-person-reid-master\torchreid\metrics\rank_cylib\rank_cy.pyx in torchreid.metrics.rank_cylib.rank_cy.evaluate_cy()
30 if use_metric_cuhk03:
31 return eval_cuhk03_cy(distmat, q_pids, g_pids, q_camids, g_camids, max_rank)
---> 32 return eval_market1501_cy(distmat, q_pids, g_pids, q_camids, g_camids, max_rank)
33
34

ValueError: Buffer dtype mismatch, expected 'long' but got 'long long'"

[KaiyangZhou]

any more info? like command to reproduce your error? did you make any change to the source code?

[darcyzhc]

any more info? like command to reproduce your error? did you make any change to the source code?

Hello KaiyangZhou. I encountered the same problem. The result is:

......
epoch: [10/60][790/808] time 0.520 (0.500)      data 0.000 (0.007)      eta 5:36:41     loss 2.1012 (1.8828)    acc 75.0000 (81.7168)   lr 0.000300
epoch: [10/60][800/808] time 0.490 (0.500)      data 0.000 (0.007)      eta 5:36:34     loss 1.8767 (1.8839)    acc 87.5000 (81.6875)   lr 0.000300
##### Evaluating market1501 (source) #####
Extracting features from query set ...
Done, obtained 3368-by-2048 matrix
Extracting features from gallery set ...
Done, obtained 15913-by-2048 matrix
Speed: 0.0383 sec/batch
Computing distance matrix with metric=euclidean ...
Computing CMC and mAP ...
Traceback (most recent call last):
  File ".\reid_test.py", line 56, in <module>
    run()
  File ".\reid_test.py", line 52, in run
    test_only=False
  File "e:\python_project\deep-person-reid\torchreid\engine\engine.py", line 211, in run
    ranks=ranks
  File "e:\python_project\deep-person-reid\torchreid\engine\engine.py", line 344, in test
    rerank=rerank
  File "D:\Anaconda3\envs\torchreid\lib\site-packages\torch\autograd\grad_mode.py", line 43, in decorate_no_grad
    return func(*args, **kwargs)
  File "e:\python_project\deep-person-reid\torchreid\engine\engine.py", line 419, in _evaluate
    use_metric_cuhk03=use_metric_cuhk03
  File "e:\python_project\deep-person-reid\torchreid\metrics\rank.py", line 201, in evaluate_rank
    use_metric_cuhk03
  File "torchreid\metrics\rank_cylib\rank_cy.pyx", line 24, in torchreid.metrics.rank_cylib.rank_cy.evaluate_cy
  File "torchreid\metrics\rank_cylib\rank_cy.pyx", line 32, in torchreid.metrics.rank_cylib.rank_cy.evaluate_cy
ValueError: Buffer dtype mismatch, expected 'long' but got 'long long'

I didn't change the source code. I just do as the guide says, no additional command. My running code is:

import torchreid
import torch

def run():
    torch.multiprocessing.freeze_support()
    print('loop')
    datamanager = torchreid.data.ImageDataManager(
        root='F:\\Data_sets\\reid-data',
        sources='market1501',
        targets='market1501',
        height=256,
        width=128,
        batch_size_train=16,
        batch_size_test=16,
        transforms=['random_flip', 'random_crop']
    )

    model = torchreid.models.build_model(
        name='resnet50',
        num_classes=datamanager.num_train_pids,
        loss='softmax',
        pretrained=True
    )

    model = model.cuda()

    optimizer = torchreid.optim.build_optimizer(
        model,
        optim='adam',
        lr=0.0003
    )

    scheduler = torchreid.optim.build_lr_scheduler(
        optimizer,
        lr_scheduler='single_step',
        stepsize=20
    )

    engine = torchreid.engine.ImageSoftmaxEngine(
        datamanager,
        model,
        optimizer=optimizer,
        scheduler=scheduler,
        label_smooth=True
    )

    engine.run(
        save_dir='log/resnet50',
        max_epoch=60,
        eval_freq=10,
        print_freq=10,
        test_only=False
    )

if __name__ == '__main__':
    run()

My system is Windows10, pytorch 1.1.0, cudatoolkit 9.0. Because Windows system ,I have to change some code compared to the get-started-30-seconds, or there will be error described as #issues21114.
And the unchanged code runs properly on my ubuntu system.
Could you help to solve this problem?

[KaiyangZhou]

seems to be a problem for windows

I can't reproduce the error without a windows machine, maybe @berkdenizi has solved the issue?

[berkdenizi]

I solved but idk how . i played on some codes long time ago .. @KaiyangZhou @darcyzhc

[darcyzhc]

I solved but idk how . i played on some codes long time ago .. @KaiyangZhou @darcyzhc

Thanks for your reply, berkdenizi. Could you remember how did you solve this problem?

[s2244521]

I also have this problem and my environment is windows. @berkdenizi , How did you solve it?

[darcyzhc]

I figure out a workaround about this problem. You can prohibit the use of cython by change attribute use_cython=True to use_cython=False in function evaluate_rank in file torchreid\metrics\rank.py. Though it'll be slower, it can work.

[darcyzhc]

And I get the workable solution now. As described in #160:

1. change every "long" to "long long" in the file /metrics/rank_cylib/rank_cy.pyx.
2. delete the file 'rank_cy.cp37-win_amd64.pyd' in the same folder manually.
3. in terminal, re-run the command 'python setup.py develop'. Then you will see a new file ''rank_cy.cp37-win_amd64.pyd' generated under folder /metrics/rank_cylib.