Open msukoz opened 7 years ago
I see that in faster rcnn paper, author uses coco model to test VOC dataset, and the mAP is high. Now, I want to use VOC and my own dataset to finetune such a caffemodel. I download from the link below, but there is a mismatch error in layer "cls_score". https://dl.dropboxusercontent.com/s/cotx0y81zvbbhnt/coco_vgg16_faster_rcnn_final.caffemodel?dl=0
I see that in faster rcnn paper, author uses coco model to test VOC dataset, and the mAP is high. Now, I want to use VOC and my own dataset to finetune such a caffemodel. I download from the link below, but there is a mismatch error in layer "cls_score". https://dl.dropboxusercontent.com/s/cotx0y81zvbbhnt/coco_vgg16_faster_rcnn_final.caffemodel?dl=0