Giters

BichenWuUCB / squeezeDet

A tensorflow implementation for SqueezeDet, a convolutional neural network for object detection.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Using negative samples for training.

bayesian-mind opened this issue · comments

commented

Hello @BichenWuUCB , I am training the Squeezedet model for object detection, and I was wondering if there is anyway I can use negative samples while training to improve accuracy. CUrrently when I use negative samples the losses start going to nan. Have you tried training using negative samples, if yes can you provide more details on the same?