Research Article

U-DAVIS-Deep Learning Based Arm Venous Image Segmentation Technique for Venipuncture

Table 2

Comparison of different augmentation methods on the basis of their PSNR, IoU, and dice scores.

PSNRIoUDice
Training with:MeanStdMinMaxMeanStdMinMaxMeanStdMinMax

No augmentations0.4350.1460.0680.6780.6820.010.5520.9610.3970.110.0670.517
AHE0.5820.1650.0450.8270.7880.0070.6590.9730.480.1410.0450.708
CLAHE0.6240.1650.0970.8860.790.0050.7180.980.5450.1460.0970.799
Both AHE & CLAHE0.7510.1550.1170.930.8930.0040.8210.9960.6850.1490.1170.871