performance

[pioneer158]

trafficstars

I use demo.py to detect a image,but its performace is bad.

[avBuffer]

Sorry! demo.py is only example, you should train your training data by using yolov5 or yolov4 or yolov3 etc. The accuracy and performance are all decided by model structure. You maybe modify yolov5 model structure to speed up. In fact, I found the performance is poor in yolov5, I'm trying to modify it. Waiting for good news!

[pioneer158]

thank you! nice work

------------------ 原始邮件 ------------------ 发件人: "Jay Meng"<[email protected]>; 发送时间: 2020年7月6日(星期一) 下午5:16 收件人: "avBuffer/Yolov5_tf"<[email protected]>; 抄送: "调得一手好参"<[email protected]>; "Author"<[email protected]>; 主题: Re: [avBuffer/Yolov5_tf] performance (#1)

Sorry! demo.py is only example, you should train your training data by using yolov5 or yolov4 or yolov3 etc. The accuracy and performance are all decided by model structure. You maybe modify yolov5 model structure to speed up. In fact, I found the performance is poor in yolov5, I'm trying to modify it. Waiting for good news!

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.