Vision UI was provided by Meituan-Vision-Testing tool.
There is no need to train model for this project，Vision-ml is an RNN training project
Setup
# create venv and activate
git clone git@github.com:Meituan-Dianping/vision-ui.git --depth=1
cd vision-ui
pip3 install -r requirements.txt
# set working directory for command-line usage
export PYTHONPATH=$PYTHONPATH:$path/to/project/vision-ui-
beyond pixel diff-vision diff
-
template match-image merge
-
pre-trained model-UI detection
-
vision semantic-semantic target recognition
-
pre-trained model-OCR
| App1 | App2 | App3 |
|---|---|---|
 |
 |
 |
| image or text query | proposal backend | result |
|---|---|---|
 |
ui-infer |  |
| a toy dinosaur | patches |  |
| mario | patches |  |
This project is licensed under the MIT