search

ria-com / nomeroff-net

Nomeroff Net. Automatic numberplate recognition system.
GNU General Public License v3.0
459 stars 159 forks source link

RuntimeError: view size is not compatible with input tensor's size and stride (at least one dimension spans across two contiguous subspaces). Use .reshape(...) instead #113

Closed aleksandr294 closed 3 years ago

aleksandr294 commented 4 years ago

2020-11-03 06:47:27.453771: W tensorflow/stream_executor/platform/default/dso_loader.cc:59] Could not load dynamic library 'cudart64_101.dll'; dlerror: cudart64_101.dll not found 2020-11-03 06:47:27.454030: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine. 2020-11-03 06:47:29.386906: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x1c9168a3260 initialized for platform Host (this does not guarantee that XLA will be used). Devices: 2020-11-03 06:47:29.387188: I tensorflow/compiler/xla/service/service.cc:176] StreamExecutor device (0): Host, Default Version 2020-11-03 06:47:29.388538: W tensorflow/stream_executor/platform/default/dso_loader.cc:59] Could not load dynamic library 'nvcuda.dll'; dlerror: nvcuda.dll not found 2020-11-03 06:47:29.388779: W tensorflow/stream_executor/cuda/cuda_driver.cc:312] failed call to cuInit: UNKNOWN ERROR (303) 2020-11-03 06:47:29.392588: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:169] retrieving CUDA diagnostic information for host: DESKTOP-2M99F75 2020-11-03 06:47:29.392850: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:176] hostname: DESKTOP-2M99F75 Config 'C:\Users\balas\PycharmProjects\pythonProject\nomeroff-net./NomeroffNet/configs/centermask2/numberplates/cpu_centermask_numberplate_V_39_eSE_FPN_ms_3x.yaml' has no VERSION. Assuming it to be compatible with latest v2. Traceback (most recent call last): File "C:/Users/balas/PycharmProjects/pythonProject/main.py", line 42, in cv_imgs_masks = nnet.detect_mask([img]) File "C:\Users\balas\PycharmProjects\pythonProject\nomeroff-net\NomeroffNet\DetectronDetector.py", line 75, in detect_mask outputs = self.predictor(im) File "C:\Users\balas\PycharmProjects\pythonProject\venv\lib\site-packages\detectron2\engine\defaults.py", line 223, in call predictions = self.model([inputs])[0] File "C:\Users\balas\PycharmProjects\pythonProject\venv\lib\site-packages\torch\nn\modules\module.py", line 727, in _call_impl result = self.forward(*input, *kwargs) File "C:\Users\balas\PycharmProjects\pythonProject\venv\lib\site-packages\detectron2\modeling\meta_arch\rcnn.py", line 149, in forward return self.inference(batched_inputs) File "C:\Users\balas\PycharmProjects\pythonProject\venv\lib\site-packages\detectron2\modeling\metaarch\rcnn.py", line 202, in inference proposals, = self.proposal_generator(images, features, None) File "C:\Users\balas\PycharmProjects\pythonProject\venv\lib\site-packages\torch\nn\modules\module.py", line 727, in _call_impl result = self.forward(input, **kwargs) File "C:\Users\balas\PycharmProjects\pythonProject\nomeroff-net\centermask2\centermask\modeling\fcos\fcos.py", line 117, in forward proposals = outputs.predict_proposals() File "C:\Users\balas\PycharmProjects\pythonProject\nomeroff-net\centermask2\centermask\modeling\fcos\fcos_outputs.py", line 362, in predict_proposals l, o, r, c, self.image_sizes File "C:\Users\balas\PycharmProjects\pythonProject\nomeroff-net\centermask2\centermask\modeling\fcos\fcos_outputs.py", line 390, in forward_for_single_feature_map pre_nms_top_n = candidate_inds.view(N, -1).sum(1)

Запуск происходит без GPU. Буду очень благодарен в решении данной проблемы.

ApelSYN commented 4 years ago

Отключите работу с CUDA с помощью переменной окружения: Для этого замените этот код 

os.environ["CUDA_VISIBLE_DEVICES"] = "0" 
os.environ["TF_FORCE_GPU_ALLOW_GROWTH"]="true"

этим

os.environ["CUDA_VISIBLE_DEVICES"] = ""
aleksandr294 commented 4 years ago

Спасибо.

vogoriachko commented 4 years ago

@ApelSYN при запуске кода на Google Colab возникает такая же ошибка и на CPU и на GPU, то есть os.environ["CUDA_VISIBLE_DEVICES"] = "" не решает проблемы. Ошибка: RuntimeError: view size is not compatible with input tensor's size and stride (at least one dimension spans across two contiguous subspaces). Use .reshape(...) instead.

ApelSYN commented 4 years ago

Ошибка может возникать из-за некорректной версии torch и torchvision, убедитесь в том что torch версии 1.6, torchvision версии 0.7.0, вот так можно все поправить:

pip3 install torch==1.6
pip3 install torchvision==0.7.0
ApelSYN commented 3 years ago

Автор centermask2 рофиксил проблему совместимости -- должно работать со свежими версичми torch и torchvision pip3 install "torch>=1.6" pip3 install "torchvision>=0.7"

Переоткройте это issue если проблема еще не решена.