The current windows version 2.6.8 of the app can be downloaded from here

Instructions for the old version of the camera review app are here

Checkout this video to learn about the updated APP and how to use it

About the app

This is the first version of the auto classification model and it is not perfect

The purpose of this app it to present an easy to use tool to categorize camera photos and create training data to identify targets in game camera photographs

We are working on a pipline to better train the model.

We are always looking for feedback ideas and ways to partner on the app

TODO

  • Get more training data to classify more targets and improve classification

  • Improve the speed of classification

  • Add better filtering

  • Detach the app from our infastructure so it is standalone

How it works

We first use microsoft's megadetctor as and object detector

We then classifiy the objects using our tensorflow model. The model we use runs on your local machine using CPU or GPU

Requirements

GPU Driver Requirements

The app can use CPU (very slow) or GPU (much faster) to auto classify images

If you use CPU you should not need to install anything and it should run in the browser

Want to speed it up?

To use GPU your machine must have the required hardware and drivers for tensorflow.js

https://github.com/tensorflow/tfjs/blob/master/tfjs-node/README.md#installing

For GPU support, the app requires the following NVIDIA® software installed on your system:

Windows / Mac OS X Requires Python 2.7

Windows & OSX build support for node-gyp requires Python 2.7. Be sure to have this version before installing

Download the app

the current windows version of the app is here

Follow these instrutions for the regular camera review work flow

Download the models

Settings -> Download -> Place them where you can find them

Create a review and set up the classification

Filter and Review Bounding boxes

Edit Bounding Boxes

Complete the review and send us your photos and annotation files to integrate into the next version of the model

Did this answer your question?