Skip to content

Latest commit

 

History

History
29 lines (24 loc) · 826 Bytes

README.md

File metadata and controls

29 lines (24 loc) · 826 Bytes

tchrs-opencv-webcam-inference

This example shows steps for running a Python trained model on webcam feed with opencv and tch-rs. Model will run on GPU.

Model used in example is mobilenet_v3_small. It is trained on Cifar10 dataset for 10 epochs. Python training and testing scripts are included in:

/train_simple_classifier

To run example below lines must be added under [dependencies] in Cargo.toml

[dependencies]
anyhow = "1.0.71"
opencv = "0.82.1"
tch = "0.13.0"

main.rs is commented for readers to understand easily.

Pretrained model and a test video also included in repo

Usage

To run example on a video use below command:

cargo run cifar10_mobilenet_v3_small.pt test_video.mp4

To run example on your webcam use below command:

cargo run cifar10_mobilenet_v3_small.pt /dev/video0