Stereo-Matching Deep Learning Dataset for Maize 3D Phenotyping

Abstract

The dataset consists of stereo images of maize plants captured daily from August 20 to September 6, aimed at constructing a comprehensive dataset for phenotypic analysis and the development of stereo vision models. The images were acquired at two resolutions: 1280×640 and 640×360 pixels. For each resolution, six individual plants were recorded, organized into 20 stereo pairs per plant, resulting in a total of 240 images per resolution. Each stereo pair comprises one image from the left camera and one from the right camera, enabling three-dimensional reconstructions and comparative growth analyses. The 640×360 images correspond to consecutive captures of the 1280×640 images, intended to facilitate future work on devices with lower computational capacity while preserving essential scene information.

Code and data availability

公開論文であることは確認できましたが、現在の公式API・許可済み取得経路では本文を自動取得できませんでした。

No evidence-backed public reproduction asset is currently recorded.