search

chensnathan / YOLOF

You Only Look One-level Feature (YOLOF), CVPR2021, Detectron2
MIT License
271 stars 28 forks source link

"bad box: x1 larger than x2" #8

Closed x-x110 closed 3 years ago

x-x110 commented 3 years ago

In the course of my training, I made this error but it was predicting delta。in the logic, it shouldn't have occurred , whether or not I downloaded the trained weights, I would have thrown this error when I iterated around 400。

chensnathan commented 3 years ago

Hi, Could you provide more details about how you train with YOLOF?

x-x110 commented 3 years ago

I'll download the code, as you said. Since your code uses distributed training by default, and I only have one graphics card, I annotated the YoloF.py (num_foreground) and num_foreground = num_foreground * 1.0 / dis.get_world_size () inside. Start training on this basis.

Note: I use your weights to be able to test.

chensnathan commented 3 years ago

Please check here to adjust the learning rate and batch size with one GPU.

x-x110 commented 3 years ago

think you!!!!!