Retaining on ssd-mobilnet imput image

[cespedesk]

Hi, I am trying to make my model see further and I learned that it is done by increasing the input image during training. How do I change the input image recognition on the Hello AI Wold costume dataset training?

[dusty_nv]

Hi @cespedesk, I have a separate branch of pytorch-ssd called res-512 which is setup for training on larger 512x512 images. For more info, please refer to this thread:

How train jetson-inference ssd512 model

You need to change those specs to the ones from that post. I gave it a try and confirmed that it trains at 512x512 resolution. I also had to make a slight modification to the ONNX exporter script to pick up the correct resolution to export at. I pushed the changes to the res-512 branch of pytorch-ssd here: https://github.com/dusty-nv/pytorch-ssd/tree/res-512 https://github.com/dusty-nv/pytorch-ssd/commit/4ac1cfc18ac9856d247ed6a8f16a7a7de05d5364