autofocus
autofocus copied to clipboard
uptake
Deep learning computer vision for classifying wildlife in camera trap images
Resolves #110 by modifying the travis config to: - copying the trained model from AWS, - building a local image on travis from the pushed branch, - running the two...
Right now I have to push new versions of the app to Dockerhub manually. Pushing them automatically as part of Travis builds when merging into master would be more reliable.
[Megadetector](https://github.com/microsoft/CameraTraps/blob/master/megadetector.md) is an object detection model for animals in camera trap images. We could use it to infer that an image is empty if there is no bounding box and...
Using only the 2016-2017 data is very limiting because it is only from mid-summer. I wouldn't expect models trained on just this data to generalize to other times of year,...
Rather than providing entire images to a classifier, it might work better to identify areas of interest as a first step and apply the classifier within those areas.
Feel free to open a PR before all of these items have been completed. Pull Request Checklist - [ ] Pull request includes a description of the change and the...
The code in `autofocus/predict` works, but it was written in a hurry and would probably benefit from some careful attention.
As per my understanding, the current interface provides a list of animals along with probability scores, which would be bit difficult for an end user to consume. A simple webpage...
README is clear in terms of the steps, but could look excessively long for new users. In order to de clutter README, it would be good to create an example...