search

BichenWuUCB / squeezeDet

A tensorflow implementation for SqueezeDet, a convolutional neural network for object detection.
BSD 2-Clause "Simplified" License
739 stars 306 forks source link

Training getting stuck at 10000 Steps #88

Closed amanmeetgarg closed 6 years ago

amanmeetgarg commented 6 years ago

Hi Group, I am trying to train the Squeezedet network for a new dataset. The training parameters have not been changed, through model, and config have been modified to accommodate the new data.

The training run is getting stuck at 10000 steps, where as my max_steps=1000000 . I tried three different runs with data augmentation on/off, new data subset for train set. to no avail.

has anybody faced similar issue to find a resolution? any insights would be valuable.

Thanks & Regards Aman

amanmeetgarg commented 6 years ago

I modified my data into the KITTI data format, and placed images in the same directory structure as well.