Distortion-Adaptive Grape Bunch Counting for Omnidirectional Images

TL;DR

This paper introduces a novel distortion-adaptive counting method for omnidirectional images, specifically applied to grape bunch counting, utilizing stereographic projection and new data augmentation techniques to improve accuracy.

Contribution

It presents the first object counting approach for omnidirectional images using stereographic projection, along with a distortion-adaptive Gaussian kernel and data augmentation, advancing the field of omnidirectional image analysis.

Findings

Improved counting accuracy with 14.7% lower MAE.

Enhanced method outperforms conventional approaches.

Constructed a new grape-bunch omnidirectional image dataset.

Abstract

This paper proposes the first object counting method for omnidirectional images. Because conventional object counting methods cannot handle the distortion of omnidirectional images, we propose to process them using stereographic projection, which enables conventional methods to obtain a good approximation of the density function. However, the images obtained by stereographic projection are still distorted. Hence, to manage this distortion, we propose two methods. One is a new data augmentation method designed for the stereographic projection of omnidirectional images. The other is a distortion-adaptive Gaussian kernel that generates a density map ground truth while taking into account the distortion of stereographic projection. Using the counting of grape bunches as a case study, we constructed an original grape-bunch image dataset consisting of omnidirectional images and conducted experiments to evaluate the proposed method. The results show that the proposed method performs better than a direct application of the conventional method, improving mean absolute error by 14.7% and mean squared error by 10.5%.