train: WARNING: Ignoring corrupted image and/or label xxxx : cannot reshape array of size 55 into shape (2)

[Rookte]

Hi, I am a beginner at this YOLOv7 pose. When i download the coco2017labels-keypoint.zip and try to train yolov7 pose with coco2017 data .i got this error.

train: WARNING: Ignoring corrupted image and/or label xxxx : cannot reshape array of size 55 into shape (2)

i cannot get a correct cache. the cache don't have any information. thanks.

[holger-prause]

did you picked the right model config - but also this is just a wild guess - i have not trained such a model before.

[Rookte]

thanks, i know how to solve the issue.

[holger-prause]

Great :-)

What was the problem actually in case someone has the same or a similar problem?

[aoqiangma]

thanks, i know how to solve the issue.

Hello, can you tell me how you solved this problem?

thanks, i know how to solve the issue.

[jq9898]

thanks, i know how to solve the issue.

Hello, I also encountered this problem. Can you tell me how you solved this problem?

[aoqiangma]

Wow, that's great. How did you solve it? Can you tell me something about it? Thank you very much.

	@.***

@.*** |

---- Replied Message ---- | From | @.> | | Date | 11/17/2022 13:26 | | To | @.> | | Cc | @.**@.> | | Subject | Re: [WongKinYiu/yolov7] train: WARNING: Ignoring corrupted image and/or label xxxx : cannot reshape array of size 55 into shape (2) (Issue #1054) |

thanks, i know how to solve the issue.

Hello, I also encountered this problem. Can you tell me how you solved this problem?

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>

[aoqiangma]

Please help me, thank you very much.

At 2022-11-17 13:26:11, "jq9898" @.***> wrote:

thanks, i know how to solve the issue.

Hello, I also encountered this problem. Can you tell me how you solved this problem?

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>

[Rookte]

Sorry, didn't see the message in time. This issue is because the parameter kpt-label is not set well. I was trained by command

python train.py --data data/coco_kpts.yaml --cfg cfg/yolov7-w6-pose.yaml --weights weights/yolov7-w6-person.pt --kpt-label

The parameter kpt-labe is very important. Otherwise, it becomes training object instances instead of object keypoints.

[jq9898]

Sorry, didn't see the message in time. This issue is because the parameter kpt-label is not set well. I was trained by command

python train.py --data data/coco_kpts.yaml --cfg cfg/yolov7-w6-pose.yaml --weights weights/yolov7-w6-person.pt --kpt-label

The parameter kpt-labe is very important. Otherwise, it becomes training object instances instead of object keypoints.

Thank you very much. This really solved my problem. During debugging, I found that the kpt-label was always false. I was still thinking about how to modify the kpt-label, and I received your email. Thank you for your help in solving my problem.

[aoqiangma]

Thanks, I'll try it this afternoon.

	@.***

@.*** |

---- Replied Message ---- | From | @.> | | Date | 11/18/2022 11:05 | | To | @.> | | Cc | @.**@.> | | Subject | Re: [WongKinYiu/yolov7] train: WARNING: Ignoring corrupted image and/or label xxxx : cannot reshape array of size 55 into shape (2) (Issue #1054) |

Sorry, didn't see the message in time. This issue is because the parameter kpt-label is not set well. I was trained by command

python train.py --data data/coco_kpts.yaml --cfg cfg/yolov7-w6-pose.yaml --weights weights/yolov7-w6-person.pt --kpt-label

The parameter kpt-labe is very important. Otherwise, it becomes training object instances instead of object keypoints.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>

[aoqiangma]

Thank you very much, I'll try it this afternoon.

	@.***

@.*** |

---- Replied Message ---- | From | @.> | | Date | 11/18/2022 11:05 | | To | @.> | | Cc | @.**@.> | | Subject | Re: [WongKinYiu/yolov7] train: WARNING: Ignoring corrupted image and/or label xxxx : cannot reshape array of size 55 into shape (2) (Issue #1054) |

Sorry, didn't see the message in time. This issue is because the parameter kpt-label is not set well. I was trained by command

python train.py --data data/coco_kpts.yaml --cfg cfg/yolov7-w6-pose.yaml --weights weights/yolov7-w6-person.pt --kpt-label

The parameter kpt-labe is very important. Otherwise, it becomes training object instances instead of object keypoints.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>

[aoqiangma]

WARNING: Dataset not found, nonexistent paths: ['F:\code\coco\coco2017\val2017\val2017.txt'] Traceback (most recent call last): File "train.py", line 562, in train(hyp, opt, device, tb_writer) File "train.py", line 97, in train check_dataset(data_dict) # check File "F:\code\yolov7-pose\utils\general.py", line 183, in check_dataset raise Exception('Dataset not found.') Exception: Dataset not found.

Thank you for your help. I followed your instructions, but the following error occurred again. However, I can't debug errors yet. I suspect it's a path or data set problem. Excuse me, is your data set all the data sets in coco128? I just selected a few data sets.

At 2022-11-18 11:05:01, "Rookte" @.***> wrote:

Sorry, didn't see the message in time. This issue is because the parameter kpt-label is not set well. I was trained by command

python train.py --data data/coco_kpts.yaml --cfg cfg/yolov7-w6-pose.yaml --weights weights/yolov7-w6-person.pt --kpt-label

The parameter kpt-labe is very important. Otherwise, it becomes training object instances instead of object keypoints.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>

[aoqiangma]

WARNING: Dataset not found, nonexistent paths: ['F:\code\coco\coco2017\val2017\val2017.txt'] Traceback (most recent call last): File "train.py", line 562, in train(hyp, opt, device, tb_writer) File "train.py", line 97, in train check_dataset(data_dict) # check File "F:\code\yolov7-pose\utils\general.py", line 183, in check_dataset raise Exception('Dataset not found.') Exception: Dataset not found.

Thank you for your help. I followed your instructions, but the following error occurred again. However, I can't debug errors yet. I suspect it's a path or data set problem. Excuse me, is your data set all the data sets in coco128? I just selected a few data sets.

At 2022-11-18 11:11:43, "jq9898" @.***> wrote:

Sorry, didn't see the message in time. This issue is because the parameter kpt-label is not set well. I was trained by command

python train.py --data data/coco_kpts.yaml --cfg cfg/yolov7-w6-pose.yaml --weights weights/yolov7-w6-person.pt --kpt-label

The parameter kpt-labe is very important. Otherwise, it becomes training object instances instead of object keypoints.

Thank you very much. This really solved my problem. During debugging, I found that the kpt-label was always false. I was still thinking about how to modify the kpt-label, and I received your email. Thank you for your help in solving my problem.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>

[aoqiangma]

Thank you for your help. I will conduct the next experiment according to your help.

At 2022-11-18 16:25:17, "马奥强" @.***> wrote:

WARNING: Dataset not found, nonexistent paths: ['F:\code\coco\coco2017\val2017\val2017.txt'] Traceback (most recent call last): File "train.py", line 562, in train(hyp, opt, device, tb_writer) File "train.py", line 97, in train check_dataset(data_dict) # check File "F:\code\yolov7-pose\utils\general.py", line 183, in check_dataset raise Exception('Dataset not found.') Exception: Dataset not found.

Thank you for your help. I followed your instructions, but the following error occurred again. However, I can't debug errors yet. I suspect it's a path or data set problem. Excuse me, is your data set all the data sets in coco128? I just selected a few data sets.

At 2022-11-18 11:11:43, "jq9898" @.***> wrote:

Sorry, didn't see the message in time. This issue is because the parameter kpt-label is not set well. I was trained by command

python train.py --data data/coco_kpts.yaml --cfg cfg/yolov7-w6-pose.yaml --weights weights/yolov7-w6-person.pt --kpt-label

The parameter kpt-labe is very important. Otherwise, it becomes training object instances instead of object keypoints.

Thank you very much. This really solved my problem. During debugging, I found that the kpt-label was always false. I was still thinking about how to modify the kpt-label, and I received your email. Thank you for your help in solving my problem.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>

[Rookte]

my data is set all the data sets in coco.

[battulaharish2]

(yolov7) C:\Users\Z0172725\yolov7>python train.py --data data/coco_kpts.yaml --img 640 --batch 32 --device cpu --weights yolov7-w6-pose.pt --name yolov7_640_val_p --kpt-label --epochs 1

usage: train.py [-h] [--weights WEIGHTS] [--cfg CFG] [--data DATA] [--hyp HYP] [--epochs EPOCHS] [--batch-size BATCH_SIZE] [--img-size IMG_SIZE [IMG_SIZE ...]] [--rect] [--resume [RESUME]] [--nosave] [--notest] [--noautoanchor] [--evolve] [--bucket BUCKET] [--cache-images] [--image-weights] [--device DEVICE] [--multi-scale] [--single-cls] [--adam] [--sync-bn] [--local_rank LOCAL_RANK] [--workers WORKERS] [--project PROJECT] [--entity ENTITY] [--name NAME] [--exist-ok] [--quad] [--linear-lr] [--label-smoothing LABEL_SMOOTHING] [--upload_dataset] [--bbox_interval BBOX_INTERVAL] [--save_period SAVE_PERIOD] [--artifact_alias ARTIFACT_ALIAS] [--freeze FREEZE [FREEZE ...]] [--v5-metric]

train.py: error: unrecognized arguments: --kpt-label

@Rookte please help in this, how to resolve this. Both for testing and training it's showing like this.

[Rookte]

（约洛夫7）C：\Users\Z0172725\yolov7>python train.py --data data/coco_kpts.yaml --img 640 --batch 32 --device cpu --weights yolov7-w6-pose.pt --name yolov7_640_val_p --kpt-label --epochs 1

用法： train.py [-h] [--权重权重] [--cfg CFG] [--数据数据] [--催眠 HYP] [--纪元纪元] [--批次大小 BATCH_SIZE] [--img-size IMG_SIZE [IMG_SIZE ...]][--右][--简历 [简历]] [--nosave] [--notest] [--noautoanchor] [--evolve] [--bucket BUCKET] [--cache-images] [--image-weights] [--device DEVICE] [--multi-scale] [--single-cls] [--adam] [--sync-bn] [--local_rank LOCAL_RANK] [--workers WORKERS] [--项目项目] [--实体实体] [--名称名称] [--exist-ok] [--quad] [--linear-lr] [--标签平滑LABEL_SMOOTHING] [--upload_dataset][--bbox_interval BBOX_INTERVAL][--save_period SAVE_PERIOD][--artifact_alias ARTIFACT_ALIAS][--冻结冻结 冻结 [冻结...]][--v5 指标]

train.py：错误：无法识别的参数：--kpt-label

@Rookte请帮助解决这个问题，如何解决这个问题。无论是对于测试和训练，它都像这样显示。

did you picked the right code - but also this is just a wild guess

[battulaharish2]

@Rookte Your guess was right, I used wrong code, later I used pose repo code. https://github.com/WongKinYiu/yolov7/tree/pose My bad, forgot to update here, thanks for the reply.