Hi! On converted caffe model estimation accurate is so good, but speed is slow for me. I wrote script for runing estimation on video like camera_pose_demo.py and with gtx1050 it shows 2fps. Can I somehow improve speed without accuracy worsing? Maybe somebody can give his owntrained model, or give me some recomendations? Thanks!
Hi! On converted caffe model estimation accurate is so good, but speed is slow for me. I wrote script for runing estimation on video like camera_pose_demo.py and with gtx1050 it shows 2fps. Can I somehow improve speed without accuracy worsing? Maybe somebody can give his owntrained model, or give me some recomendations? Thanks!