Dataset - Maize Ear Sensing Dataset for On-Farm Yield Predictions

This dataset contains RGB + depth sensor data for maize ear yield prediction collected during field trials at Kansas. The data was captured using an Intel RealSense depth camera and includes synchronized image streams stored in ROS bag format, along with corresponding ground truth measurements of ear morphology and grain yield. The dataset supports the paper "Maize ear sensing for on-farm yield predictions" presented at the CVPR 2025 Workshop on Vision for Agriculture. It enables research on nondestructive crop yield estimation using depth sensing and computer vision techniques. Contents:- ROS bag files (.bag) with synchronized RGB and depth image streams from Intel RealSense camera- Ground truth CSV file (field_sample.csv) containing measurements for each sample: ear length, width, volume, fresh weight, dry weight, moisture content, kernel count, and kernel dry weight- Two sample groups with distinct identifiers for cross-validation Total size: ~6 GBGrowth stage: R4 and R6 (physiological maturity) For usage instructions and code to extract images from bag files, see the accompanying GitHub repository: https://github.com/Ciampitti-Lab/MaizeEarSensing