YouEye - kiosk machine helper solution for blinded people

Using no-voice supported kiosk for blinded people is is extremly difficult.

Therefore, if you just reach your hand and capture it with your phone camera, YouEye will read you what will you touch.

Currently, only python demos are available, with web-server wrapper.
Word coordinate detector for https://github.com/code-yeongyu/east-detector-socket-wrapper is used for this project.

Current works

Demonstration Video

Demonstration Pictures

https://raw.githubusercontent.com/code-yeongyu/YouEye/master/python-demo/demo-images/result_0.txt

https://raw.githubusercontent.com/code-yeongyu/YouEye/master/python-demo/demo-images/result_1.txt

https://raw.githubusercontent.com/code-yeongyu/YouEye/master/python-demo/demo-images/result_2.txt

https://raw.githubusercontent.com/code-yeongyu/YouEye/master/python-demo/demo-images/result_3.txt

https://raw.githubusercontent.com/code-yeongyu/YouEye/master/python-demo/demo-images/result_4.txt

All these ocr-ed results are genereated by naver-ocr-api, both naver-ocr-api and tesseracts are available.
Because of the lack of trainset, I recommend you to use naver-ocr-api.

Run demonstration

clone this project, move to directory python-demo and install required pip modules by executing:

pip install -r requirements.txt

then execute:

python demonstration.py

For web-interface demonstration:

python web-wrapper.py

About web interface

Send psot request to /i_am_iron_man with form-data body, and attach your image to key image.
Then the server will respond with the desired text.
Currently, the web interface is setted to use naver-ocr-api, so you have to get permission or use tesseract to use ocr properly.

swk3169 / YouEye