NVIDIA TAO - detectnet_v2 - 0mAP problem

Description

Whenever I train the detectnet_v2 model using tao toolkit in visdrone dataset, I usually get 0mAP. I also checked that if this issue is because of wrong class maps but the classmaps are fine but still model doesn’t learn anything.

Environment

TensorRT Version: 7.2
GPU Type: RTX2070 SUPER
Nvidia Driver Version: 470.63.01
CUDA Version: 11.0
CUDNN Version: 8.0
Operating System + Version: ubuntu 20
Python Version (if applicable): 3.6.9
TAO version: 3.21.08
TensorFlow Version (if applicable): 1.14
Baremetal or Container (if container which image + tag): TAO container

Relevant Files

https://drive.google.com/drive/folders/1TDwJhyrzunHUCyCpmRL-3teey3n60s4G?usp=sharing

Steps To Reproduce

  1. Download the visdrone dataset and convert it to tfrecords using dataset_convert tool
  2. After converting, train the detectnet_v2 model on it, this causes a 0mAP problem and the model doesn’t learn anything

I followed the steps from the NVIDIA TAO documentation to train the detectnet_v2 model

Hi,

Looks like this is TAO related. Moving post to TAO forum to get better help.

Thank you.

I couldn’t find the forum. It would be of great help if you could share some link.

The same topic as NVIDIA TAO - detectnet_v2 - 0mAP problem. Let us track on that.

1 Like

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.