sachsbl/segmental

Dockerized Semantic Image Segmentation service using Tensorflow, Keras, and DeepLabv3+

segmental's Issues

Docker pull gives error on run -> botocore Unable to locate credentials

I followed the instructions

You can also pull my Docker image:

docker pull sachsbl/segmental
docker run -p 80:5000 segmental

Which I had to change to

docker pull sachsbl/segmental
docker run -p 80:5000 sachsbl/segmental

to work by the way. (otherwise it cannot find the image locally after the pull)

But when I executed that last command I got the following error:

botocore.exceptions.NoCredentialsError: Unable to locate credentials

how to input an image?

hello,when I input an image , I have some problem.so please tell me how to input an image to test.
Thanks,I am a recruit。

Dockerfile pip requirements.txt is giving error: No matching distribution found for tensorflow==2.0.0-beta1

When building my own docker image from the dockerfile I get this error:

ERROR: Could not find a version that satisfies the requirement tensorflow==2.0.0-beta1
ERROR: No matching distribution found for tensorflow==2.0.0-beta1

I tried removing the version (just putting 'tensorflow' in the requirements.txt) and that enabled me to succesfully built the image. But there are still some things going wrong.

To test this more easily I removed the entrypoint from the dockerfile (I didn't need the webserver anyway, just wanted an easy way to install) and then ran the image with:

sudo docker run --rm segmentallocal bash -c 'python /app/segmental_cli.py /app/test/test_images/cat.jpg'

That gives me the following error:

TypeError: Could not build a TypeSpec for <KerasTensor: shape=(None, 64, 64, 256) dtype=float32 (created by layer 'tf.compat.v1.image.resize_images')> with type KerasTensor