pytorch-yolo-v3
pytorch-yolo-v3 copied to clipboard
In my subjects,I find a problems during the person detect,that exist false retrieval.
As you see in the images,The detected object is false,How can i improving this false performance.
https://github.com/ultralytics/yolov3
Namespace(cfg='cfg/yolov3-spp.cfg', conf_thres=0.5, data_cfg='data/coco.data', images='data/samples', img_size=416, nms_thres=0.5, weights='weights/yolov3-spp.weights')
Using CPU
image 1/2 data/samples/unnamed.png: 5 persons, Done. (0.593s)
https://github.com/ultralytics/yolov3
命名空间(cfg = ' cfg / yolov3-spp.cfg ',conf_thres = 0.5,data_cfg = ' data / coco.data ',images = ' data / samples ',img_size = 416,nms_thres = 0.5,weights = ' weights / yolov3 -spp.weights ') 使用CPU 图像1 / 2数据/样品/ unnamed.png:5人,完成。(0. 593s)
thanks your reply! I change my nms_thres and own yolov3.weights,I fixed the error.I think that's overfeat problems.Because iterate 40000.