Performance Optimization Background
Watermark Removal is developed based on TensorFlow 1.15 and has some hardware requirements. It may face performance bottlenecks when running on low-configuration computers.
Specific Optimization Options
- Using Google Colab: The project supports Colab runs with free access to Google's GPU resources, completely bypassing local hardware limitations.
- Reduced processing resolution: Shrink the input image to 256×256 or smaller with the -resize parameter.
- Enable memory optimization: add tf.config.optimizer.set_jit(True) in main.py to enable XLA compilation acceleration
- batch mode: Modify the code to achieve batch processing and reduce the overhead of repeatedly loading models
alternative
If you still can't run smoothly, you can consider 1) using a more lightweight OpenCV+Numpy implementation of the base repair algorithm; 2) replacing it with a PyTorch version of a similar project; and 3) limiting the resource usage through Docker containers.
This answer comes from the articleWatermark Removal: open source image watermark removal tool, picture watermark recovery original imageThe
May not be reproduced without permission:AI productivity tools " How to optimize the efficiency of Watermark Removal Tool on low-configuration computers?
