Changed the program name to SaigeVision from srVision.
Supports use of up to three GPUs when training.
Classification/Detection/Segmentation: Added validation and Save best model functions.
Classification/Detection/Segmentation: Added options to choose the network depth.
Classification: Changed the existing Standard model to the Shallow model. Added the Standard and Deep model.
Detection: Changed the existing Standard model to the Shallow model. Added the Standard model.
Segmentation: Added the Deep model.
Improvements
Improved the Image List.
Classification: Added a function to filter images according to the label's class information in the Database Label tab.
Classification/Detection/Segmentation: Added a function to filter images according to the true and predicted label's class information in the Analyze tab.
Classification/Detection/Segmentation: Added a function to filter images according to the predicted label's class information in the Inspect tab.
Detection: Modified internal algorithms and stopped displaying IOU in the Analyze tab.
Segmentation: Adopted a new classwise threshold with values between 0 and 255 in the Analyze tab by integrating the existing overall threshold and classwise weight.
