Training Fails #72
Comments
|
@agombert Hello! It seems that the error is due to the image source file? Which image format is your using? |
|
Yes, it's what I saw, and tried to debug with openCV. But that's quite weird as it worked with the same config on a folder I had a month ago, and now not anymore. I'm using |
|
@agombert Maybe this is due to the update of albumentations? Did you update your albumentations version? |
|
I'm using the installation from this repo on a conda environment with |
|
@JulioZhao97 I tested with different versions of albumentations and got the same problem. That's quite weird. I looked into other potential libraries but only albumentations got quite recent updates. I saw this issue that may be related to it, but unsure as I'm not an expert in cv. I can give you a sample of data for you to test id you'd like to test it yourself. |
|
@agombert Can you use following setting and see whether this problem will be solved? |
|
@agombert Hello! Could you please provide your sample data? You can upload it here or send it via my email: [email protected] |
|
Hey @JulioZhao97 👋 I've just sent you the email with a link to the data sample. @hengrui0516, thanks for your help 🙏, I tried to go with the versions you mentioned (even in the extra of the pyproject.toml) but unfortunately it did not work either. |
|
@agombert I will see to it today |
|
@agombert It turns out that I can train with your sample data successfully config: task: detect
mode: train
model: yolov10m-doclayout.yaml
data: data.yaml
epochs: 500
time: null
patience: 100
batch: 1
imgsz: 1120
save: true
save_period: 10
val_period: 1
cache: false
device: '3'
workers: 4
project: public_dataset/data
name: yolov10m-doclayout_data_epoch500_imgsz1120_bs1_pretrain_None
exist_ok: false
pretrained: true
optimizer: SGD
verbose: true
seed: 0
deterministic: true
single_cls: false
rect: false
cos_lr: false
close_mosaic: 10
resume: null
amp: true
fraction: 1.0
profile: false
freeze: null
multi_scale: false
overlap_mask: true
mask_ratio: 4
dropout: 0.0
val: true
split: val
save_json: false
save_hybrid: false
conf: null
iou: 0.7
max_det: 300
half: false
dnn: false
plots: true
source: null
vid_stride: 1
stream_buffer: false
visualize: false
augment: false
agnostic_nms: false
classes: null
retina_masks: false
embed: null
show: false
save_frames: false
save_txt: false
save_conf: false
save_crop: false
show_labels: true
show_conf: true
show_boxes: true
line_width: null
format: torchscript
keras: false
optimize: false
int8: false
dynamic: false
simplify: false
opset: null
workspace: 4
nms: false
lr0: 0.02
lrf: 0.01
momentum: 0.9
weight_decay: 0.0005
warmup_epochs: 3.0
warmup_momentum: 0.8
warmup_bias_lr: 0.1
box: 7.5
cls: 0.5
dfl: 1.5
pose: 12.0
kobj: 1.0
label_smoothing: 0.0
nbs: 64
hsv_h: 0.015
hsv_s: 0.7
hsv_v: 0.4
degrees: 0.0
translate: 0.1
scale: 0.5
shear: 0.0
perspective: 0.0
flipud: 0.0
fliplr: 0.5
bgr: 0.0
mosaic: 1.0
mixup: 0.0
copy_paste: 0.0
auto_augment: randaugment
erasing: 0.4
crop_fraction: 1.0
cfg: null
tracker: botsort.yaml
save_dir: public_dataset/data/yolov10m-doclayout_data_epoch500_imgsz1120_bs1_pretrain_NoneI provide my environment for your reference: |
|
Hey @JulioZhao97 thanks for your help. I'll try a couple of things to see if I can handle the problem and let you know asap ! |
|
Ok @JulioZhao97 I found the problem !! 😌 It was coming from a 🐍 |
|
Thank you very much for your help ! 🙏 |
Hey got an error in training which is quite weird when doing:
Looks like there is a problem to read the picture with openCV, but I tried with an old dataset I used to train the model and same error. Do you have any idea what's happening ?
Best,
Arnault
The text was updated successfully, but these errors were encountered: