I am trying to run ssd mobilenetv2 model on jetson nano for object detection. I needed help with a couple of basic queries.
- How many labels is the model trained on? Is there a label.txt file with the list of labels
- What dataset is this model (the one that is pre-installed on jetson nano) trained on? Right now, I am getting inconsistent object detections. Is there a way to improve the accuracy of the model?
Thanks,
Tauseef
Hi @tauseefahmad12, if you mean the one from jetson-inference, it is trained on the 90-class MS COCO dataset. The labels.txt file can be found here: https://github.com/dusty-nv/jetson-inference/blob/master/data/networks/ssd_coco_labels.txt
To improve the detections, you could increase the confidence threshold with the --threshold
parameter, or collect data and train your own model with https://github.com/dusty-nv/jetson-inference/blob/master/docs/pytorch-collect-detection.md