#vision #tensorflow #object #detection #computer

aperature

Object detection using single shot multibox detector

2 releases

✓ Uses Rust 2018 edition

0.1.1 Jan 26, 2019
0.1.0 Jan 15, 2019

#50 in Machine learning

Download history 12/week @ 2019-01-12 4/week @ 2019-01-19

7 downloads per month

MIT license

14KB
163 lines

aperature

Build Status Docs crates.io

Alt text

A Rust crate for doing object detection using Tensorflow!

aperature uses the ssd_mobilenet developed by Google for object detection. ssd_mobilenet provides a quick execution time as well as a small size to allow for the most diverse range of deployment of the crate.

Documentation

Usage

Input: Input images can be provided through img::GenericImage that can be found in this crate or by passing in the Image struct found in the pison image crate. To use other image types not listed here there is a public trait DetectionImage that is also accepted as input to the model.

Output: Arrays containing the detection infomration found in the image. This information includes the bounding boxes of found objects, the confidence score for those found objects, the detection classes for the objects, and the number of detections in the image.

Changelog

Project changes and releases will be captured in the CHANGELOG.md file that can be found in this repository.

Code of Conduct

The aperature project follows a strict community code of conduct. Information about our code of conduct can be found at CODE_OF_CONDUCT.md

Contact

Mason Stallmo masonstallmo@gmail.com

Dependencies

~9.5MB
~166K SLoC