search

talebolano / yolov3-network-slimming

yolov3 network slimming剪枝的一种实现
344 stars 93 forks source link

training error! #2

Open lucheng07082221 opened 6 years ago

lucheng07082221 commented 6 years ago

Hi: 我在训练的时候到这里报错：

loss_cls = (1 / nB) * self.ce_loss(pred_cls[mask], torch.argmax(tcls[mask], dim=1))

Traceback (most recent call last): File "/home/lc/work/yolov3-network-slimming/sparsity_train.py", line 159, in train() File "/home/lc/work/yolov3-network-slimming/sparsity_train.py", line 107, in train loss = model(imgs, targets) File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py", line 477, in call result = self.forward(*input, kwargs) File "/home/lc/work/yolov3-network-slimming/yolomodel.py", line 365, in forward x, losses = self.module_list[i][0](x, targets) File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py", line 477, in call result = self.forward(input, kwargs) File "/home/lc/work/yolov3-network-slimming/yolomodel.py", line 147, in forward print(torch.argmax(tcls[mask], dim=1)) File "/usr/local/lib/python3.6/dist-packages/torch/functional.py", line 374, in argmax return torch._argmax(input, dim, keepdim) RuntimeError: Dimension out of range (expected to be in range of [-1, 0], but got 1)

talebolano commented 6 years ago

hi， 请问是在一开始就报错还是在训练中报错，这有可能是训练集标签不匹配造成的 我的yolo源码来自于https://github.com/eriklindernoren/PyTorch-YOLOv3 貌似有人也遇到了相似的问题

lucheng07082221 commented 6 years ago

@talebolano 这个问题解决了，就是没找到标签，label和images要在同一个目录下面，谢谢哈！

lucheng07082221 commented 6 years ago

@talebolano 这个可以多块gpu一起训练吗

talebolano commented 6 years ago

@talebolano 这个可以多块gpu一起训练吗

目前还不可以，我会在近期添加多块gpu运行的代码

Liqing6668 commented 5 years ago

@talebolano 您好，我在训练的时候报错，请您帮忙看一下。 [Epoch 0/2000, Batch 155/4402] [Losses: x 0.184448, y 0.152233, w 0.473080, h 0.243340, conf 0.419103, cls 1.269924, total 2.742128, recall: 0.59055, precision: 0.05823] [Epoch 0/2000, Batch 156/4402] [Losses: x 0.161208, y 0.169744, w 0.484731, h 0.234620, conf 0.445205, cls 1.283632, total 2.779141, recall: 0.61207, precision: 0.05506] [Epoch 0/2000, Batch 157/4402] [Losses: x 0.180584, y 0.142397, w 0.394638, h 0.239581, conf 0.393050, cls 1.264419, total 2.614670, recall: 0.61757, precision: 0.06406] [Epoch 0/2000, Batch 158/4402] [Losses: x 0.168928, y 0.124468, w 0.472078, h 0.200339, conf 0.371993, cls 1.261448, total 2.599255, recall: 0.61994, precision: 0.05572] [Epoch 0/2000, Batch 159/4402] [Losses: x 0.156252, y 0.146104, w 0.336951, h 0.280383, conf 0.460892, cls 1.268392, total 2.648975, recall: 0.60119, precision: 0.05369] [Epoch 0/2000, Batch 160/4402] [Losses: x 0.186453, y 0.147905, w 0.433254, h 0.227833, conf 0.377308, cls 1.252797, total 2.625550, recall: 0.60784, precision: 0.08115] [Epoch 0/2000, Batch 161/4402] [Losses: x 0.181762, y 0.173695, w 0.988128, h 0.325833, conf 0.431874, cls 1.305756, total 3.407048, recall: 0.43548, precision: 0.05132] [Epoch 0/2000, Batch 162/4402] [Losses: x 0.174461, y 0.164167, w 0.488938, h 0.336979, conf 0.619493, cls 1.290267, total 3.074305, recall: 0.52143, precision: 0.05007] [Epoch 0/2000, Batch 163/4402] [Losses: x 0.209021, y 0.128292, w 0.700684, h 0.202287, conf 0.369960, cls 1.244245, total 2.854489, recall: 0.56198, precision: 0.04331] [Epoch 0/2000, Batch 164/4402] [Losses: x 0.216554, y 0.121116, w 0.389169, h 0.224613, conf 0.367337, cls 1.234381, total 2.553170, recall: 0.58206, precision: 0.07246] [Epoch 0/2000, Batch 165/4402] [Losses: x 0.202888, y 0.139989, w 0.332888, h 0.200512, conf 0.407665, cls 1.273087, total 2.557030, recall: 0.58230, precision: 0.06962] [Epoch 0/2000, Batch 166/4402] [Losses: x 0.164832, y 0.102894, w 0.821020, h 0.175726, conf 0.359270, cls 1.234681, total 2.858424, recall: 0.60841, precision: 0.05036] [Epoch 0/2000, Batch 167/4402] [Losses: x 0.171323, y 0.160461, w 0.659578, h 0.529196, conf 0.378674, cls 1.251355, total 3.150587, recall: 0.55263, precision: 0.05115] [Epoch 0/2000, Batch 168/4402] [Losses: x 0.178747, y 0.127256, w 0.795630, h 0.410524, conf 0.363543, cls 1.255784, total 3.131485, recall: 0.53846, precision: 0.03526] Traceback (most recent call last): File "sparsity_train.py", line 153, in train() File "sparsity_train.py", line 100, in train loss = model(imgs, targets) File "/opt/anaconda3/lib/python3.6/site-packages/torch/nn/modules/module.py", line 492, in call result = self.forward(*input, kwargs) File "/home/Liqing/Liqing/yolov3-network-slimming-master/yolomodel.py", line 349, in forward x, losses = self.module_list[i][0](x, targets) File "/opt/anaconda3/lib/python3.6/site-packages/torch/nn/modules/module.py", line 492, in call result = self.forward(input, kwargs) File "/home/Liqing/Liqing/yolov3-network-slimming-master/yolomodel.py", line 102, in forward img_dim=self.image_dim, File "/home/Liqing/Liqing/yolov3-network-slimming-master/util.py", line 187, in build_targets conf_mask[b, anch_ious > ignore_thres, gj, gi] = 0 IndexError: index 21 is out of bounds for dimension 3 with size 17

liaoyunkun commented 5 years ago

@lucheng07082221 您好，我在pascal voc数据集上遇到了这个问题，请问您是在那个数据集上测试遇到这个问题，具体是如何解决？

pandasong commented 5 years ago

@Liqing6668 您好，请问下IndexError: index 21 is out of bounds for dimension 3 with size 17这个问题您解决了吗，是怎么解决的？我也遇到同样的问题！非常感谢！！！