AssertionError: Checkpoint not found. analogous to #237 issue

[DomenicoMessina]

Hi,

I've encountered the same issue of #237.
I tried to run this google colab notebook, without positive results

https://colab.research.google.com/drive/1Kpy4QGFZ2ZHm69mPfkmLSUes8kj6Bjyi?usp=sharing#scrollTo=79ib1d3xFpAy

You developed a fantastic library. I hope you will fix this issue.
Thanks,
Domenico

[divamgupta]

I just ran colab this notebook. And it works perfectly fine and I am not able to reproduce the error you are facing.

[DomenicoMessina]

Thank you for your fast answer, Divamgupta.

I've added to the notebook the module google.colab to have the access to drive, using the following code

import os
from google.colab import drive
drive.mount('/gdrive/',force_remount = True)

os.symlink('/gdrive/My Drive/', 'my_drive')

After that, I manually downloaded and extracted your dataset in a subfolder of my_drive folder. Maybe the code do not recognize this?

[PaulaScharf]

I also have the same error. It runs fine when I run the google colab but if I try the same commands on my home PC it cant find the checkpoints. Specifically this line does not seem to find any files in my checkpoints-folder.

[divamgupta]

Which OS are you using and what is exact command you are using?

[PaulaScharf]

I am on Windows 10 home (x64).
I downloaded the datasets with the given link.
Then I ran
python -m keras_segmentation train --checkpoints_path="./checkpoints/vgg_unet_1" --train_images="./dataset1/images_prepped_train/" --train_annotations="./dataset1/annotations_prepped_train/" --val_images="./dataset1/images_prepped_test/" --val_annotations="./dataset1/annotations_prepped_test/" --n_classes=50 --input_height=320 --input_width=640 --model_name="vgg_unet" --epochs=5
and then
python -m keras_segmentation predict --checkpoints_path="./checkpoints/vgg_unet_1" --input_path="./dataset1/images_prepped_test/0016E5_07959.png" --output_path="./out24.png".
Then I get the same error as in #237.

It works if I change this line to the following:
all_checkpoint_files = glob.glob(checkpoints_path + "*.*")

[divamgupta]

Thanks. This package has not been tested with windows. How does the checkpoints directory look like? I was trying to understand why it currently fails.

…

On Wed, Mar 31, 2021 at 4:51 AM Paula Scharf ***@***.***> wrote: I am on Windows 10 home (x64). I downloaded the datasets with the given link <https://drive.google.com/file/d/0B0d9ZiqAgFkiOHR1NTJhWVJMNEU/view?usp=sharing> . Then I ran python -m keras_segmentation train --checkpoints_path="/tmp/vgg_unet_1" --train_images="dataset1/images_prepped_train/" --train_annotations="dataset1/annotations_prepped_train/" --val_images="dataset1/images_prepped_test/" --val_annotations="dataset1/annotations_prepped_test/" --n_classes=50 --input_height=320 --input_width=640 --model_name="vgg_unet" --epochs=5 and then python -m keras_segmentation predict --checkpoints_path="/tmp/vgg_unet_1" --input_path="dataset1/images_prepped_test/0016E5_07959.png" --output_path="./out24.png". Then I get the same error as in #237 <#237>. It works if I change this line <https://github.com/divamgupta/image-segmentation-keras/blob/52ec91e1625d67a79126d8cf9a287d64d03f4e9f/keras_segmentation/train.py#L21> to the following: all_checkpoint_files = glob.glob(checkpoints_path + "*.*") — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#286 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAONR5JWVLDLU3P3IM7MY73TGLPC3ANCNFSM4ZZXQYEQ> .

[PaulaScharf]

(I aborted training after 1 epoch because it takes so long)

[divamgupta]

Thanks, I have fixed it in the latest commit.