Hi. Any suggestions for training on binary detection dataset? Only one bounding box for every input image and all the labels are the same (only one object to detect). The input size is 480*640. Need specific advice on the hyper parameters such as neg_pos_rate, ssd300 vs. ssd512, ...etc. Any helps would be appreciated, thank you so much!
Hi. Any suggestions for training on binary detection dataset? Only one bounding box for every input image and all the labels are the same (only one object to detect). The input size is 480*640. Need specific advice on the hyper parameters such as neg_pos_rate, ssd300 vs. ssd512, ...etc. Any helps would be appreciated, thank you so much!