Is there a version of the Guassian YOLOv3 tiny?

[xinsuinizhuan]

My object is not complex, but precision should be achieved and speed should be fast. I use the tiny_yolos3 detect before, so whether is there a version of the Guassian YOLOv3 tiny??

[AlexeyAB]

tiny_gaus.cfg.txt

[nyj-ocean]

@AlexeyAB I found this tiny_gaus.cfg.txt is similar to yolo_v3_tiny_pan3_aa_ae_mixup_scale_giou.cfg.txt

By the way , I want to know if I can use yolo_v3_tiny_pan3_aa_ae_mixup_scale_giou.cfg.txt to train my own dataset or common dataset like VOC? My own dataset is not sequential frames from one or several videos.

[AlexeyAB]

I want to know if I can use yolo_v3_tiny_pan3_aa_ae_mixup_scale_giou.cfg.txt to train my own dataset or common dataset like VOC

Yes.

My own dataset is not sequential frames from one or several videos.

sequential frames - is required only for LSTM models ("lstm" in the name of model)

[AlexeyAB]

You just can replace [yolo] by [Gaussian_yolo] in any network, f.e.: https://github.com/AlexeyAB/darknet/blob/master/cfg/yolov3-tiny_3l.cfg

and change filters=(classes + 9)x3

[xinsuinizhuan]

You just can replace [yolo] by [Gaussian_yolo] in any network, f.e.: https://github.com/AlexeyAB/darknet/blob/master/cfg/yolov3-tiny_3l.cfg

and change filters=(classes + 9)x3

OK.Thank you very much.