pgigioli / darknet_ros

YOLO integration with ROS for real-time object detection
51 stars 23 forks source link

failed when using yolo.cfg and yolo.weights #2

Closed HolaMirai closed 7 years ago

HolaMirai commented 7 years ago

hi, pgigioli I'm very sorry to bother you,when i use the yolo-tiny.cfg and yolo-tiny.weights ,the program works;but when i change to yolo.yolocfg and yolo.weights, error occurs . that is 22: Convolutional Layer: 13 x 13 x 512 image, 1024 filters -> 13 x 13 x 1024 image 23: Convolutional Layer: 13 x 13 x 1024 image, 1024 filters -> 13 x 13 x 1024 image 24: Convolutional Layer: 13 x 13 x 1024 image, 1024 filters -> 13 x 13 x 1024 image 25: Route Layer: 16 26: Type not recognized: [reorg] Unused field: 'stride = 2' 27: Route Layer: 26 24 28: Layer before convolutional layer must output image.: File exists ROS_interface: /home/lile/catkin_ws/src/darknet_ros-master/src/utils.c:181: error: Assertion0' failed. Aborted (core dumped)`

and i can't work it out ,can you help me ? thank you your time!

pgigioli commented 7 years ago

You are using the cfg file for yolo_v2. darknet_ros is only compatible with yolo_v1 for now.