deepFruits: Datasets and annotation tool.

Please find the document below that describes how to fine-tune Faster-RCNN for fruits detection. It includes the code patch that allows fine-tuning by yourself and annotation tool (written in MATLAB). The dataset can be found from here.

Tutorial documentation

In order to perform object detection with Faster-RCNN for sweep pepper or your objects, we need to patch the original Faster-RCNN with this file. The version of py-faster-rcnn (https://github.com/rbgirshick/py-faster-rcnn) we used is;

py-faster-rcnn
commit 81ee2272c76a89748d56355a048ffce7c05a3e67
Author: Ross Girshick <rbg@fb.com>
Date:   Wed Oct 28 08:52:34 2015 -0700
Set cfg.GPU_ID in demo.py

caffe-fast-rcnn
commit 4115385deb3b907fcd428ac0ab53b694d741a3c4
Author: Ross Girshick <ross.girshick@gmail.com>
Date: Sun Aug 9 13:41:39 2015 -0700

 

Please note that the fine-tuning requires a NVIDIA CUDA GPU which has more than 8GB GPU memory. Make sure your graphics card can handle this before running.

Please also note that this software package has been verified only local HPC (high-performance computing) environments. It may be different to your Linux system environments (e.g., dependencies). But we provide troubleshooting guidance that we have experienced in the above document.

 

Please cite our related publication with this dataset and documentation if you use the proposed framework or dataset.

@Article{s16081222,
AUTHOR = {Sa, Inkyu and Ge, Zongyuan and Dayoub, Feras and Upcroft, Ben and Perez, Tristan and McCool, Chris},
TITLE = {DeepFruits: A Fruit Detection System Using Deep Neural Networks},
JOURNAL = {Sensors},
VOLUME = {16},
YEAR = {2016},
NUMBER = {8},
PAGES = {1222},
URL = {http://www.mdpi.com/1424-8220/16/8/1222},
ISSN = {1424-8220},
DOI = {10.3390/s16081222}
}

 

One thought on “deepFruits: Datasets and annotation tool.

  1. Moussa Mansour

    Dear,

    I read your paper DeepFruits: A fruit detection systems using deep neural networks and I am trying to implement the proposed model. However, I tried to download the files z_ge-py-faster-rcnn.tar and v2.4_full_screen_works.zip but it seems that the links are broken. Is there another repository to get this files?

    Thank you,
    Moussa Mansour

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>