dense_prediction_transformers

dense_prediction_transformers

Input

image file (576x384)

(Image from https://pixabay.com/ja/photos/%E5%A5%B3%E3%81%AE%E5%AD%90-%E5%98%98-%E3%82%AF%E3%83%A9%E3%82%B7%E3%83%83%E3%82%AF%E3%82%AB%E3%83%BC-1209321/)

Output

image file (576x384)

--task=segmentation

--task=monodepth

Usage

Automatically downloads the onnx and prototxt files on the first run. It is necessary to be connected to the Internet while downloading.

If you run by onnxruntime instead of ailia, you use --onnx option.

This sample has monodepth and segmentation task. You have to add --task=segmentation in the case of running segmentation task and --task=monodepth in the case of running monodepth task.

Below is example of running segmentation task by cpu.

$ python3 dense_prediction_transformers.py -i input.jpg -s output.png --task=segmentation -e 0
$ python3 dense_prediction_transformers.py -i input.jpg -s output.png--task=monodepth -e 0

After running this program, the predicted image are saved in output.png.

Reference

dense_prediction_transformers

Framework

PyTorch 1.8.1

Model Format

ONNX opset = 11

Netron

dpt_hybrid_monodepth.onnx.prototxt
dpt_hybrid_segmentation.onnx.prototxt