Renaud Bouckenooghe

I would be interested by the same functionality

Hi @MartyG-RealSense, I'm a colleague of @GEngels and @Doch88, and I'm taking over this issue. I'm currently writing a first long message to try to be the most complete possible....

I think that as pjreddie has updated the yolo.cfg file (one [local] and one [dropout] layers instead of the before last [connected] layer), he has also updated the pre-trained weights...

As said by @yhzyhzyhz, it works when installing cython 0.20 then rebuilding and installing pyvision ``` sudo pip install cython==0.20 cd pyvision sudo python setup.py build sudo python setup.py install...

Hi @genekogan, Just for info, this fork of Darknet (that is more maintained that the original one) is providing a C++ api. Nice what you're doing for Runway. Edit: https://github.com/AlexeyAB/darknet

Hi @johanjohan, Sorry, I forgot to paste the link. Here it is: https://github.com/AlexeyAB/darknet

I'm having the same error message on Linux (Ubuntu 16.04) CUDA 8 installed

I would also be interested knowing that, with MOT 2d bounding boxes format "Classification-Aware HOTA" in the paper

I would be really interested too by the inclusion of Darknet YOLO in Multitarget-tracker

I have the same problem