@@ -156,3 +156,8 @@ results of image size 608/416/320 above.
**NOTE**: MobileNet-SSD is trained in 2 GPU with totoal batch size as 64 and trained 120 epoches. VGG-SSD is trained in 4 GPU with total batch size as 32 and trained 240 epoches. SSD training data augmentations: randomly color distortion,