frkhan/bangla-ocr

Using docker to build the app instead of gradio in Huggingface spaces.

-- Dynamically set model directory based on the environment (HuggingFace / Docker).

-- Switched back to Gradio sdk for HF deployment since HF Spaces does not allow to create folder under /app folder

-- Changed the deployment stack to docker for HF.

-- Included Github Rep, Story Link, Tech Stack Badges in the UI

Switched back to Gradio SDK for HF SPaces deployment.

Merge branch 'main' of https://github.com/KI-IAN/python-ocr-webapp

Switched back to Gradio sdk since HF does not permit to create a file in it's container.

Testing docker sdk to deploy in Hugging Face Spaces.

Merge branch 'main' of https://github.com/KI-IAN/python-ocr-webapp

Last attempt: using /tmp folder.

Using Gradio SDK to deploy in HF Spaces.

Changed the path to /app/models

Changed the model location to /tmp/.EasyOCR

Using Path.home() which should be the /root folder in Linux.

Added the URL of the demo site.

Merge branch 'main' of https://github.com/KI-IAN/python-ocr-webapp

EasyOCR tries to write its model files to the root directorywhich is nto writable inside most Docker containers - especially on Huggin Face Spaces. So we explicitly set EasyOCR's model_storage_directory to a relative path inside the project directory.

Used docker to build and deploy the app in Hugging Face Space.

Merge branch 'main' of https://github.com/KI-IAN/python-ocr-webapp

Added the Configuration Header so that Hugging Face can tell Spaces how to run this app.

Workflow will be triggered on main branch now.

Create sync-to-hf action to push the commits from github repo to hugging face space repo.

Create LICENSE

Added a proper README.md file.

Enabled English (en) with Bangla (bn) script.

server_name="0.0.0.0" makes Gradio listen on all network interfaces (inside the Docker container).

Commented out the volumes mount section.

Using correct pytorch docker image.

Used pytorch official image to build docker container.

Using python:3.10 image that should include gcc by default.

Using full python-3.10 image.

Install torch first using the official PyTorch repo for CPU

Using full python image.

Removed few pip installation parameters

Added additional depencies in dockerfile.

Renamed the Dockerfile

Added the docker-compose file.

Showing time taken to extract the text.

A gradio app to convert image into text using easyocr.