search

vanstorm9 / yugioh-one-shot-learning

One shot learning (image recognition) on 10,856 unique Yugioh cards
96 stars 19 forks source link

Instructions #1

Closed maciek-jk closed 3 years ago

maciek-jk commented 3 years ago

Hello, I have trouble running this program with my own image set, would you mind including a short instruction in the readme how to run it (which files to run, etc.). Also, can you include a requirements.txt file with packages needed?

vanstorm9 commented 3 years ago

Python command instructions were added to the readme as well as adding requirements.txt to the repo.

You must have you imageset directory in this format: -cardDatabaseFull ----cardname1 --------cardname1.png ----cardname2 --------cardname2.png ----cardname3 --------cardname3.png

maciek-jk commented 3 years ago

I think something still might be missing from the requirements.txt, I get this error on running train.py ModuleNotFoundError: No module named 'models_lpf'

vanstorm9 commented 3 years ago

Ah I see, I just did a git push of the 'models_lpf' scripts into the repo. You should be able to run the script now

maciek-jk commented 3 years ago

Thanks, that fixed it, but now running to new errors. Traceback (most recent call last): File "predict.py", line 41, in <module> assert os.path.exists(loadPath) AssertionError seems its these two files that are missing hardcoded in the predict.py `loadPath = './savedModels/res-withShift-150-072020.pth'

assert os.path.exists(loadPath)

dictPath = './savedModels/featureMap-withShift-150-072020.pkl'

assert os.path.exists(dictPath) `

vanstorm9 commented 3 years ago

Download it from this link: https://drive.google.com/drive/folders/1JZCt7hHf4NYgEp2XiE1SpNboAbgBxwDY?usp=sharing