Hello, Im trying to do transfer learning on custom dataset on ssd on TAO toolkit on google colab. it seems like i can only choose resnet18 as the model (or am i wrong). i know i would have to provide my dataset on kittti format or coco format following this method here. Is it even possible to do transfer learning on google colab from TAO toolkit? or is it just to train on a that general dataset detailed in ssd.ipynb
Yes, it is possible to train on your own dataset.
do you know of any other forums that detail the process or any other documentation? I saw in the TAO documentation that i can use COCO dataset format to be able to train, would i have to change anything in the provided code on the google colab? or can i just put the dataset where the KITTI dataset should go?
There is no update from you for a period, assuming this is not an issue anymore. Hence we are closing this topic. If need further support, please open a new one. Thanks
You can prepare your own dataset with KITTI format. Then run the SSD colab or SSD notebook GPU-optimized AI, Machine Learning, & HPC Software | NVIDIA NGC
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.