search

fenglinglwb / MAT

MAT: Mask-Aware Transformer for Large Hole Image Inpainting
Other
752 stars 84 forks source link

Problems in custom dataset training #113

Open qwexacc opened 5 months ago

qwexacc commented 5 months ago

I'm using a custom dataset, which is small, 4000 + images, and running train.py after running the dataset_tool.py code which is having problems。 Total 36231618 736 - -

Setting up augmentation... Distributing across 1 GPUs... Setting up training phases... Exporting sample images... Initializing logs... Training for 2 kimg...

tick 0 kimg 0.0 time 1m 03s sec/tick 8.1 sec/kimg 4043.29 maintenance 54.8 cpumem 4.43 gpumem 24.70 augment 0.000 Evaluating metrics...

Waited a long time, been Evaluating metrics... Here, the program does not continue to run. The code I'm running is：python train.py --outdir=~/training-runs --data=~/mydataset.zip --gpus=1，Other parameters have not changed, may I ask what is the problem?

FHomeworker commented 3 months ago

Hi，I am also trying to train my own custom dataset,what should i put in my data_train file and my data_eva file?Could you please help me?A green hand ,thx.If ok,you can show me the right directory structure🥰.

danielanojan commented 2 months ago

Hi I am facing the issue. Could you get it fixed? Thank You