Apple detection

This page is devoted to our research on apple detection in industrial orchards.

Datasets

OrchardAppleDet-MSU

In our research, we use OrchardAppleDet-MSU dataset jointly collected and prepared by Bioengineering Department (Faculty of Biology), and Laboratory of Mathematical Methods of Image Processing (Faculty of Computational Mathematics and Cybernetics), Lomonosov Moscow State University.

The images depict ripe apple fruits of Gala cultivar in the commercial orchard located in Karachay-Cherkessia, Russia. The year of the orchard planting is 2017.

The dataset contains 147 images of 4000x3000px with 3363 manually segmented apple instances.

An example of the image and apple masks from the dataset:



Initial image


Apple masks

Detection method

Our apple segmentation algorithm is based on Mask-RCNN method.

The preliminary results on OrchardAppleDet-MSU dataset and Fuji-SfM dataset can be found on the following tensorboards:

Our team

Dmitry Nesterov
BC student, Laboratory of Mathematical Methods of Image Processing, Faculty of Computational Mathematics and Cybernetics, Lomonosov Moscow State University

Dmitry Sorokin
dsorokin@cs.msu.ru
ORCID: 0000-0003-3299-2545
Senior researcher, Laboratory of Mathematical Methods of Image Processing, Faculty of Computational Mathematics and Cybernetics, Lomonosov Moscow State University

Andrey Krylov
kryl@cs.msu.ru
ORCID: 0000-0001-9910-4501
Professor, Head of the Laboratory of Mathematical Methods of Image Processing, Faculty of Computational Mathematics and Cybernetics, Lomonosov Moscow State University

Andrey Nasonov
nasonov@cs.msu.ru
ORCID: 0000-0002-7844-4627
Senior researcher, Laboratory of Mathematical Methods of Image Processing, Faculty of Computational Mathematics and Cybernetics, Lomonosov Moscow State University

Alexey Solovchenko
solovchenko@mail.bio.msu.ru
ORCID: 0000-0001-6746-8511
Professor, Bioengineering Department, Faculty of Biology, Lomonosov Moscow State University

Boris Shurygin
shu_b@mail.ru
ORCID: 0000-0001-9650-3523
Bioengineering Department, Faculty of Biology, Lomonosov Moscow State University

Acknowledgement

The work is supported by the project "Intelligent Image Analysis for high-performance plant phenotyping and precision farming” of the school "Brain, cognitive systems, artificial intelligence" of Lomonosov Moscow State University.