- Released
- 27 April 2020
New Features
- Classification/Detection/Segmentation/Image Generation: Added the Continue Training function.
- Image Generation: Added an option to choose the OK : Defect ratio.
- Image Generation: Added an option to generate grayscale patches when generating defects.
- Made it possible to freely select one of the GPUs from number 0 to number 3 if multi-GPU license is available.
Improvements
- Classification/Detection/Segmentation: Subdivided the right side of the Analyze tab into the classwise result panel and overall result panel.
- Segmentation: Made it possible for users to set the Input patch size and Resize factor instead of setting the Crop size and Resize factor.
- The Crop size is calculated as 'Input patch size / Resize factor'.
Bug Fixes
- Detection: Fixed a bug that caused the difference between validation result during training and inspection result in the Analyze tab.
