Open croust opened 1 year ago
@croust it is not fixed. No PR have been merged. https://github.com/WongKinYiu/yolov7/pull/1272 is till open.
It can be solved if the numpy version numpy==1.23.5 used instead of numpy==1.24.0. I have tried it and the issue is resolved. 'pip install --upgrade numpy==1.23.5' in the project environement for pip
@suriyahgit that’s not a solution. If you do it this way, you’ll always be bound to this numpy version. We shouldn’t restrict numpy version but solve the actual problem in the code.
@suriyahgit Thanks, I have solved the issue.
I ran python train.py --workers 1 --device 0 --batch-size 8 --epochs 100 --img 640 640 --data C:\Users\mdari\YOLOv7_Custom\yolov7-custom\data\custom_data.yaml --hyp C:\Users\mdari\YOLOv7_Custom\yolov7-custom\data\hyp.scratch.custom.yaml --cfg C:\Users\mdari\YOLOv7_Custom\yolov7-custom\cfg\training\yolov7-custom.yaml --name yolov7-custom --weights yolov7.pt
and got the following error.
C:\Users\mdari\miniconda3\envs\yolov7_custom\lib\site-packages\torch\functional.py:568: UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing argument. (Triggered internally at C:\actions-runner_work\pytorch\pytorch\builder\windows\pytorch\aten\src\ATen\native\TensorShape.cpp:2228.) return _VF.meshgrid(tensors, **kwargs) # type: ignore[attr-defined] Model Summary: 415 layers, 37196556 parameters, 37196556 gradients, 105.1 GFLOPS
Transferred 552/566 items from yolov7.pt Scaled weight_decay = 0.0005 Optimizer groups: 95 .bias, 95 conv.weight, 98 other train: Scanning 'C:\Users\mdari\YOLOv7_Custom\yolov7-custom\data\train\labels.cache' images and labels... 60 found, 0 m Traceback (most recent call last): File "C:\Users\mdari\YOLOv7_Custom\yolov7-custom\train.py", line 616, in
train(hyp, opt, device, tb_writer)
File "C:\Users\mdari\YOLOv7_Custom\yolov7-custom\train.py", line 245, in train
dataloader, dataset = create_dataloader(train_path, imgsz, batch_size, gs, opt,
File "C:\Users\mdari\YOLOv7_Custom\yolov7-custom\utils\datasets.py", line 69, in create_dataloader
dataset = LoadImagesAndLabels(path, imgsz, batch_size,
File "C:\Users\mdari\YOLOv7_Custom\yolov7-custom\utils\datasets.py", line 418, in init
bi = np.floor(np.arange(n) / batch_size).astype(np.int) # batch index
File "C:\Users\mdari\miniconda3\envs\yolov7_custom\lib\site-packages\numpy__init.py", line 284, in getattr__
raise AttributeError("module {!r} has no attribute "
AttributeError: module 'numpy' has no attribute 'int'
Common solutions online says to find any files named 'numpy.py', I checked and I don't have it. I've also uninstalled and reinstalled numpy through pip.