yolact

yolact copied to clipboard

first of all, thank you for sharing code. There was a problem running the code.

root@ce7bd19200ca:/workspace/yolact# python train.py --config=yolact_base_config --batch_size=5 Multiple GPUs detected! Turning off JIT. loading annotations into memory... Done (t=11.95s) creating index... index created! loading annotations into memory... Done (t=1.59s) creating index... index created! Initializing weights... Begin training!

Traceback (most recent call last): File "train.py", line 382, in train() File "train.py", line 257, in train losses = criterion(out, wrapper, wrapper.make_mask()) File "/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py", line 493, in call result = self.forward(*input, **kwargs) File "/opt/conda/lib/python3.6/site-packages/torch/nn/parallel/data_parallel.py", line 152, in forward outputs = self.parallel_apply(replicas, inputs, kwargs) File "/opt/conda/lib/python3.6/site-packages/torch/nn/parallel/data_parallel.py", line 162, in parallel_apply return parallel_apply(replicas, inputs, kwargs, self.device_ids[:len(replicas)]) File "/opt/conda/lib/python3.6/site-packages/torch/nn/parallel/parallel_apply.py", line 83, in parallel_apply raise output File "/opt/conda/lib/python3.6/site-packages/torch/nn/parallel/parallel_apply.py", line 59, in _worker output = module(*input, **kwargs) File "/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py", line 493, in call result = self.forward(*input, **kwargs) File "/workspace/yolact/layers/modules/multibox_loss.py", line 141, in forward pos_idx = pos.unsqueeze(pos.dim()).expand_as(loc_data) RuntimeError: The expanded size of the tensor (19248) must match the existing size (14436) at non-singleton dimension 1. Target sizes: [2, 19248, 4]. Tensor sizes: [2, 14436, 1]

what should I do?

Not sure if this is the issue, but if you're using multiple GPUs, you should use a batch size that's evenly divisible by the number of GPUs you're using. The first line says you're using multiple GPUs, so let me know if that works.

Hi @dbolya, im experiencing this as well, it happens only when I try to set the pred_aspect_ratios to have different number of pred_aspect_ratio at each scales as follow: 'pred_aspect_ratios': [ [[1]], [[1/2, 2]], [[1/2, 2]], [[1]] ] is setting like this not allowed?

same to me , I change make_priors in multibox_loss.py , now it works!