Skip to content

An experiment to re-purpose MTCNN for other uses than facial detection

License

Notifications You must be signed in to change notification settings

caleb221/MTCNN-Leaf

Repository files navigation

MTCNN-Leaf

An experiment to re-purpose MTCNN for other uses than facial detection This repo is the source code for implementing MTCNN using Caffe.

Introduction

This was the first half of my Thesis project. The goal is to re-purpose the MTCNN Facial detection model and use it on an ESP-32 AI-Thinker WiFi Camera module. --> That code is in another repository: ESP32-Leaf

MTCNN

https://arxiv.org/pdf/1604.02878.pdf

SETUP

You'll need caffe installed, and setup can be done in the same way as

https://github.com/lincolnhard/mtcnn-head-detection

There are a few Data preprocessing python scripts I also made to help clean out the datasets I used. They are included at the top level directory.

There is also a python script that extracts the weights from a Caffe model and forms them into a numpy array file. This is used to translate the weights from one framework (caffe) to another (ESP-Face) The translation code is included in the ESP32-Leaf repository because it is a specific translation for that framework.

Output

--> could use some improvement, but overall it found some leaves

Input Test P-Net Output R-Net Output O-Net Output

Data Sets

There were 2 datasets used to train the model Found here

Phenotype Data set

https://www.plant-phenotyping.org/datasets-overview

100 Plant leaves

https://archive.ics.uci.edu/ml/datasets/One-hundred+plant+species+leaves+data+set

References

The code that this half of the project was based off is found here

https://github.com/lincolnhard/mtcnn-head-detection

Special thanks to Lincolnhard and the owners of the Data sets

M. Minervini, A. Fischbach, H.Scharr, and S.A. Tsaftaris. Finely-grained annotated datasets for image-based plant phenotyping. Pattern Recognition

Charles Mallah, James Cope, James Orwell. Plant Leaf Classification Using Probabilistic Integration of Shape, Texture and Margin Features. Signal Processing, Pattern Recognition and Applications, in press. 2013.

About

An experiment to re-purpose MTCNN for other uses than facial detection

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published