search

SKRohit / Improving-YOLOv3

Few training heuristics and small architectural changes that can significantly improve YOLOv3 performance with tiny increase in inference cost.
GNU General Public License v3.0
12 stars 7 forks source link

RuntimeError: CUDA error: device-side assert triggered #2

Open EslamHaourn opened 3 years ago

EslamHaourn commented 3 years ago

Hi, I face this error with the lines: Traceback (most recent call last): File "train.py", line 172, in loss = model(imgs, targets) File "/home/eslam/anaconda3/lib/python3.8/site-packages/torch/nn/modules/module.py", line 727, in _call_impl result = self.forward(*input, *kwargs) File "/home/eslam/Desktop/PyTorch-YOLOv3-kitti-master/model.py", line 294, in forward x, layer_loss = module[0](x, targets, img_dim) File "/home/eslam/anaconda3/lib/python3.8/site-packages/torch/nn/modules/module.py", line 727, in _call_impl result = self.forward(input, **kwargs) File "/home/eslam/Desktop/PyTorch-YOLOv3-kitti-master/model.py", line 188, in forward iou_scores, class_mask, obj_mask, noobj_mask, tx, ty, tw, th, tcls, tconf = build_targets( File "/home/eslam/Desktop/PyTorch-YOLOv3-kitti-master/utils/utils.py", line 469, in build_targets noobj_mask[b[i], anchor_ious > ignore_thres, gj[i], gi[i]] = 0 RuntimeError: CUDA error: device-side assert triggered

what should I do? Thanks in advance.

EslamHaourn commented 3 years ago

I tried: elif not self.darknet and module_def["type"] == "yolo":

Train phase: get loss

            if is_training:
                x, *losses = module[0](x, targets, img_dim)
                for name, loss in zip(self.loss_names, losses):
                    self.losses[name] += loss
            # Test phase: Get detections
            else:
                x = module(x)

but gave the same error