Revision history [back]

Hello, I am trying to perform object detection by using Yolov3 cfg and weights using readNetFromDarknet(cfg_file, weight_file). I run into an opencv issue as the layer_type = 'shortcut' is missing from the opencv implementation of Yolov2. Unlike layer_type = 'route' in Yolov2, shortcut has linear activation as well. So, it cannot be simply replaced and opencv recompiled. Until the recent Yolov3 change trickles down to opencv, is there any possible workaround that anyone can think of? Thanks in advance!

Hello, I am trying to perform object detection by using Yolov3 cfg and weights using readNetFromDarknet(cfg_file, weight_file). I run into an opencv issue as the layer_type = 'shortcut' is missing from the opencv implementation of Yolov2. Unlike layer_type = 'route' in Yolov2, shortcut has linear activation as well. So, it cannot be simply replaced and opencv recompiled. Until the recent Yolov3 change trickles down to opencv, is there any possible workaround that anyone can think of? Thanks in advance!

Hello, I am trying to perform object detection using Yolov3 cfg and weights using via readNetFromDarknet(cfg_file, weight_file). weight_file) in opencv. I run into an opencv issue as the layer_type = 'shortcut' is missing from the opencv implementation of Yolov2. Unlike layer_type = 'route' in Yolov2, shortcut has linear activation as well. So, it cannot be simply replaced and opencv recompiled. Until the recent Yolov3 change trickles down to opencv, is there any possible workaround that anyone can think of? Thanks in advance!