TypeError: 'NoneType' object is not subscriptable

[yhx0105]

Have there anyone else who met the same problem as I met?
It can train in some batch,however meet the error suddenly.
I am sure the path is correct.but I don't know how to solve it
---- [Epoch 0/100, Batch 7/55] ---- +------------+--------------+--------------+--------------+ | Metrics | YOLO Layer 0 | YOLO Layer 1 | YOLO Layer 2 | +------------+--------------+--------------+--------------+ | grid_size | 16 | 32 | 64 | | loss | 53.503120 | 49.941383 | 60.964851 | | x | 0.109302 | 0.079437 | 0.094549 | | y | 0.089693 | 0.093862 | 0.099689 | | w | 1.409947 | 0.991294 | 2.199120 | | h | 0.381467 | 0.345485 | 4.021410 | | conf | 50.943905 | 47.888214 | 53.957447 | | cls | 0.568806 | 0.543090 | 0.592637 | | cls_acc | 2.04% | 0.00% | 12.24% | | recall50 | 0.000000 | 0.000000 | 0.000000 | | recall75 | 0.000000 | 0.000000 | 0.000000 | | precision | 0.000000 | 0.000000 | 0.000000 | | conf_obj | 0.380452 | 0.323845 | 0.408052 | | conf_noobj | 0.388892 | 0.365904 | 0.407760 | +------------+--------------+--------------+--------------+

File "train.py", line 100, in <module> for batch_i, (_, imgs, targets) in enumerate(dataloader): File "/root/anaconda2/envs/keyan/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 568, in __next__ return self._process_next_batch(batch) File "/root/anaconda2/envs/keyan/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 608, in _process_next_batch raise batch.exc_type(batch.exc_msg) TypeError: Traceback (most recent call last): File "/root/anaconda2/envs/keyan/lib/python3.6/site-packages/torch/utils/data/_utils/worker.py", line 99, in _worker_loop samples = collate_fn([dataset[i] for i in batch_indices]) File "/root/anaconda2/envs/keyan/lib/python3.6/site-packages/torch/utils/data/_utils/worker.py", line 99, in <listcomp> samples = collate_fn([dataset[i] for i in batch_indices]) File "/home/yanghanxun/PyTorch-YOLOv3_05/utils/datasets.py", line 130, in __getitem__ img, targets = horisontal_flip(img, targets) File "/home/yanghanxun/PyTorch-YOLOv3_05/utils/augmentations.py", line 8, in horisontal_flip targets[:, 2] = 1 - targets[:, 2] TypeError: 'NoneType' object is not subscriptable

[xiaoxiaoshu111]

Hi,I also met the same problem,Have you solved it?
Thanks~

[tyjcalvert]

I ran into this issue and it ended up being caused by having training images that did not have corresponding labels. I removed the images that did not have annotations and that fixed the problem. Hope that fixes the issue for you as well.

[yhx0105]

I ran into this issue and it ended up being caused by having training images that did not have corresponding labels. I removed the images that did not have annotations and that fixed the problem. Hope that fixes the issue for you as well.

oh ,thanks .Maybe I have a bad data set

[yhx0105]

Hi,I also met the same problem,Have you solved it?
Thanks~

你解决这个问题了吗？ 我在datasets.py第100行把
if self.augment: 改为：
if self.augment and targets is not None:
就可以正常运行了，希望对你有帮助

[Shiny-ZhangXiXin]

It could be this: your image suffix being uppercase JPG or uppercase PNG causes problems with dataset reading your label_files.You can do it this way:
path.replace("images", "labels").replace(".png", ".txt").replace(".jpg", ".txt")
-->path.replace("images", "labels").replace(".PNG", ".txt").replace(".JPG", ".txt")
Check carefully to see if your label files have been read successfully by debugging

[Note-Liu]

Hi,I also met the same problem,Have you solved it?
Thanks~

你解决这个问题了吗？ 我在datasets.py第100行把
if self.augment: 改为：
if self.augment and targets is not None:
就可以正常运行了，希望对你有帮助

other error appear：
/PyTorch-YOLOv3-master/utils/datasets.py", line 142, in collate_fn
targets = torch.cat(targets, 0)
RuntimeError: expected a non-empty list of Tensors

[kravrolens]

Hi,I also met the same problem,Have you solved it?
Thanks~

你解决这个问题了吗？ 我在datasets.py第100行把
if self.augment: 改为：
if self.augment and targets is not None:
就可以正常运行了，希望对你有帮助

灰常感谢！

[kravrolens]

I ran into this issue and it ended up being caused by having training images that did not have corresponding labels. I removed the images that did not have annotations and that fixed the problem. Hope that fixes the issue for you as well.

It is True, Thank you!

[luisete2]

I checked my images and every single one of them has annotations. Removing them is no use, I tried that. I checked that all the suffixes are lowercase too, and I'm unable to run this. Any ideas?

[juagvera]

Revisé mis imágenes y cada una de ellas tiene anotaciones. Eliminarlos no sirve de nada, lo intenté. Verifiqué que todos los sufijos también están en minúsculas y no puedo ejecutar esto. ¿Algunas ideas?

could you solve it?

[chaishazi]

楼主解决了这个问题了吗，我现在也遇到了这个问题，不知道怎么弄了

[Flova]

This part of the code has been updated or removed. I, therefore, close this issue now. Feel free to reopen for further discussion.

[JaGuzmanT]

I ran into this issue and it ended up being caused by having training images that did not have corresponding labels. I removed the images that did not have annotations and that fixed the problem. Hope that fixes the issue for you as well.

Thank you very much. I tried with half of my dataset in order to get out the incomplete or erroneous labels and I finally could identify the labels with problems. Greets!!