Supplementary Materials for: Robustness Enhancement of Self-Localization for Drone-View Mixed Reality via Adaptive RGB-Thermal Integration

This dataset contains the supplementary materials, including the Unity project source code, experimental result videos, and quantitative measurement data, for the manuscript submitted to MDPI Technologies (Special Issue: Image Analysis and Processing). 1. Contents Unity Project: Source code for the proposed RGB-Thermal fusion simulation. Core Script: OnlineLocalization.cs implements the "Effective Inlier Count ($N_{eff}$)" algorithm using OpenCV. Video Folder: Contains input footage (Routes A–C) and output result videos under various conditions. Data Folder: Contains result.xlsx with detailed time-series IoU measurement results. 2. Important Notes on Running the Unity Project Please read carefully: The provided Unity project contains the complete implementation of the proposed algorithm. However, due to security and licensing reasons, specific credentials and third-party assets are excluded. You must configure the project with your own environment to run the simulation. A. Third-Party Asset Requirement OpenCV for Unity: The script OnlineLocalization.cs relies on "OpenCV for Unity" to calculate the spatial distribution of feature points. Due to licensing restrictions, this asset is NOT included. Action: Please import your own copy of OpenCV for Unity from the Unity Asset Store. B. Immersal VPS Configuration (Map Data) The system uses Immersal Cloud Service for localization. You need to prepare your own map data for the 2D-3D point matching. Generate Token & Map: Sign up at the [Immersal Developer Portal] and get your Developer Token. Scan your environment using the Immersal Mapper App and upload it to the server. Note your Map ID. Prepare 3D Point Cloud (JSON): Download the map data as a PLY file from the Immersal Developer Portal. Convert the PLY file into a JSON file (compatible with the project's format). This JSON data serves as the 3D points for the 2D-3D matching algorithm. Configure Inspector: In the Unity Inspector for OnlineLocalization: Enter your Developer Token and Map ID. Assign your converted JSON file to the "Map Json File" field. C. Input Video Configuration (DJI Mavic 3T) Prerequisite: This project is designed to process dual-screen video footage (RGB + Thermal side-by-side) captured by a DJI Mavic 3T. Action: Prepare your own video file recorded with a DJI Mavic 3T (or equivalent dual-stream format). Assign your video file to the Video Input field in the Inspector. By following these steps, the system will apply the proposed Adaptive RGB-Thermal Integration logic to your environment and video data. 3. License Code & Data: Creative Commons Attribution 4.0 International (CC-BY 4.0) Third-Party Assets: All rights to third-party assets (Immersal SDK, OpenCV for Unity) belong to their respective owners. Funding:This research was partially funded by JSPS KAKENHI, Grant Number 23K11724.