About the mAP
liguilan1227 opened this issue · 1 comments
liguilan1227 commented
Good code as you did. When I tried to obey this code, some error maybe occured. I use the PascalVoc07+12 dataset, catch the model_v1.py, but the results shows the mAP was 0.287. Is it usual?
cannguyen275 commented
I have the same interest as @liguilan1227 . Using Ghostnet in classification task archive very good results.
But how about it when applying to object detection tasks, especially SSD?