This repository contains swift scripts and Core ML models for our real-time spider recognition app
A video demo of our app: https://www.youtube.com/watch?v=LvMvKccfGkA&t=2s
Link to the deep learning model development repo: https://github.com/zhenyy/poisonous-spider-recognition
Offers two functions:
- Classifying a single spider by taking a photo or supplying a photo from library
- Real-time classification from camera feed
- Display both spider type and toxicity
- Locating and classifying multiple spiders by taking a photo or supplying a photo from library
- Real-time detection from camera feed
- Display spider types, toxicity and bounding boxes indicate spider locations
A gallery contains detailed introductions of different spiders
A map showing the locations of the spiders shared by users
Link to the Spider Map repo: https://github.com/zhenyy/SpiderMap
-
The project is not possible without this nice series of tutorials created by pyimagesearch:
How to (quickly) build a deep learning image dataset
Keras and Convolutional Neural Networks (CNNs)
Running Keras models on iOS with CoreML -
And this blogpost written by Matthijs Hollemans:
MobileNetV2 + SSDLite with Core ML
with Source code