Design and Development of System for Fruit Detection and Counting

DOI

10.2991/978-94-6239-628-9_26How to use a DOI?

Keywords

Fruit Classification; Detection; Mobile Sensor; Raspberry pi; Faster R-CNN

Abstract

India is one of the world’s major exporters of agricultural produce, particularly fruits, vegetables, and other crop varieties. With the further increase of the population of the country, the need to determine the estimates of the crop production and quality is becoming more significant. This study presents an automated method of identifying fruits and the rough yield prediction using deep learning techniques. Its primary objective is to develop a quick, precise, and reliable method of detecting fruits and using it as a key component of autonomous agricultural systems. This kind of system is also essential in predicting yields as well as its application in automated systems of harvesting. Fruit detection and classification is through observation of visual features such as color, shape and size. The project commences with the image-processing techniques and the traditional machine learning classifiers and then develops to the deep neural network techniques. The fruit detection model is developed by using a Faster Region-Based Convolutional Neural Network (Faster R-CNN). The training is performed with the help of the TensorFlow Object Detection framework, the training one takes place with the help of the set of RGB (color) and Near-Infrared (NIR) images. Multimodal training method is used to incorporate the use of both RGB and NIR features that improves convergence and increases detection and classification performance. This results in a multimodal Faster R-CNN model which is more accurate than the previous models. This system significantly increases accuracy and recall and reduces the time taken to detect apples by 40-50 seconds to approximately 0.2 seconds. In addition to better accuracy, the practical advantages of this solution are also: this method does not require pixel-level annotations; and instead, only bounding-box annotations are required, which are far less time-intensive to create. Also, a regression-based methodology is presented to enumerate the count of fruits in clusters or bunches, and thus help in making more accurate yield predictions.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Abhishek Kumar Saxena
AU  - Mritunjay Rai
AU  - Gaurav Verma
AU  - Anjali Singh
AU  - Pavan Mishra
AU  - Jatin Gaur
PY  - 2026
DA  - 2026/03/31
TI  - Design and Development of System for Fruit Detection and Counting
BT  - Proceedings of the International Conference on Recent Trends in Intelligent Computing, Manufacturing, and Electronics (rTIME 2025)
PB  - Atlantis Press
SP  - 287
EP  - 294
SN  - 2352-5401
UR  - https://doi.org/10.2991/978-94-6239-628-9_26
DO  - 10.2991/978-94-6239-628-9_26
ID  - Saxena2026
ER  -