twang comments

Results 51 comments of


                                            twang

Multi-GPU Training 🌟

hi! @glenn-jocher for multi-gpu training, if using smaller batch size than 64, could you suggest the hyperparameter to adjust like the learning rate?

Train Custom Data Tutorial ⭐

@glenn-jocher yolov5 is really flexible for training on custom datasets! could you please share the script for processing and converting coco to the train2017.txt file? I already have this file...

when distilling Faster RCNN with FPN, do we need to firstly assign gt box to different FPN level and then generate imitation region? Or for every FPN level, we generate imitation region according to all gt boxes?

on each level of fpn, only the assigned gt box will be used calculate the imitation region

can't load model!

student 是 VGG11，所以student 需要 imagenet pretrained vgg11

can't load model!

可以连接一下vpn吧？ On Sat, Jan 2, 2021 at 4:24 PM successhaha wrote: > student 是 VGG11，所以student 需要 imagenet pretrained vgg11 > > 博主能发个百度云的与训练权重吗？国内下不了 > > — > You are receiving this...

when distillation with FPN，which feature map shuold we select to calculate imitation loss, feature map of FPN(P3、P4...) or feature of backbone(C3、C4...) or others?

Hi, fpn features (p3, p4...) are used.

when distillation with FPN，which feature map shuold we select to calculate imitation loss, feature map of FPN(P3、P4...) or feature of backbone(C3、C4...) or others?

on each level of fpn, only the assigned gt box will be used calculate the imitation region

twang

Multi-GPU Training 🌟

Train Custom Data Tutorial ⭐

when distilling Faster RCNN with FPN, do we need to firstly assign gt box to different FPN level and then generate imitation region? Or for every FPN level, we generate imitation region according to all gt boxes?

can't load model!

can't load model!

when distillation with FPN，which feature map shuold we select to calculate imitation loss, feature map of FPN(P3、P4...) or feature of backbone(C3、C4...) or others?

when distillation with FPN，which feature map shuold we select to calculate imitation loss, feature map of FPN(P3、P4...) or feature of backbone(C3、C4...) or others?

warm step

warm step

vgg11蒸馏跟非蒸馏的模型测试结果是一样的