基于Paddle+Flask的眼部医疗辅助系统, 本项目基于PaddleX提供的FastSCNN语义分割模型,在眼部图像视盘分割数据集上进行训练,并开发了前后端分离项目。
后端代码基于Flask开发,前端WEB界面基于VUE开发。
cd frontend
npm install
npm run build
# npm run serve
cp -r dist/* ../backend/static
cd ..
virtualenv .venv
pip install -r requirements.txt
python app.py
使用模型进行预测,同时使用pdx.seg.visualize
将结果可视化,可视化结果将保存到./output/deeplab
下,其中weight
代表原图的权重,即mask可视化结果与原图权重因子。
import paddlex as pdx
model = pdx.deploy.Predictor('inference_model')
image_name = 'optic_disc_seg/JPEGImages/H0005.jpg'
result = model.predict(image_name)
pdx.seg.visualize(image_name, result, weight=0.4, save_dir='./output/deeplab')
2021-01-23 08:16:45 [INFO] The visualized result is saved as ./output/deeplab/visualize_H0005.jpg
!zip -r inference_model/ weights.zip
zip warning: name not matched: weights.zip
zip error: Nothing to do! (try: zip -r inference_model/ . -i weights.zip)