Full Description
This book constitutes the refereed proceedings of the 53rd Applied Imagery Pattern Recognition Workshop, AIPR 2025, held in Washington, DC, USA, during October 13-14, 2025.
The 41 papers included in this volume were carefully reviewed and selected from 60 submissions. They focus on Imagery-Related challenges in Generative AI and Large Language Models, highlighting the accelerating impact of generative technologies on imagery, video, and multimodal reasoning.
Contents
Image-Based Geospatial Foundation Models: What's the Catch?.-Can AI Generated Items Be Trusted?.-Synthetic to Street: Generative AI-Powered Object Detection of License-Plate Jackets.-Agentic Learning with an AI Instructor in Virtual Reality for High School Anatomy Instruction.-Generative Edit Agent for Retrieval & Synthesis.-An Immersive AI-Driven Virtual Reality Training for Accessible Agricultural Education in a Unity-Based VR Environment.-Examination of How an Adversarial or Naive Change to an Image May Be Tracked Visually, Layer by Layer, Using Associative Memory Matrices in Vector Form.-Vision-Language Integration for Image Captioning Using Vision Transformers and GPT-J.-Task Prioritization for Remote Sensing AI Foundation Models.-Scaling Remote Sensing Foundation Models: Data Domain Tradeoffs at the Peta-Scale.-Cross-Modal Foundation Models for Remote Sensing.-Unsupervised Change Detection and Categorization with Remote Sensing Foundation Models.-Assumptions Implicit in Applied Imagery Pattern Recognition.-AgriGen: a Prompt-Tuned, Multilingual LLM-Based Q&A System for Smarter Agriculture.-PREFUSE: a Probabilistic Reliability-Weighted Fusion for Weather-Aware Perception in Autonomous Vehicles.-Hier-SimCLR-Drive: Learning a Two-Level Traffic-Scene Taxonomy Without Labels for Autonomous Driving.-Scaling Continuous Kernels with Sparse Fourier Domain Learning.-Benchmarking Defense Techniques for Securing Large Language Models.-Advancing Foundation Models with Geospatial and Temporal Reasoning in Multi-Modal Applications.-Training the Right ML Model and Training the ML Model Right.-A Multimodal IoT-Based Smart Desk System for Real-Time Thermal Comfort Classification in Educational Environments.-Visual Feature Tracking Algorithm for Veterinary Lameness Assessment in Horses.-A Merge-Aware Metric and Boundary-Guided Head: Mergers Reduction and Improved Building Segmentation in Remote Sensing Images.-Pixel-Point Fusion: A 2D-3D Computer Vision Framework for Robust Sidewalk Trip Hazard and Crack Detection.-Neurophysiological Study of EEG Alpha-Beta Dynamics in Temporal and Parietal Regions During Non-Native and Neutral Music Listening.-Fusion of 2D LiDAR and Vision-Based Detection for Collision-Aware Indoor Navigation.-Dark Channel Prior Infused All-in-One Dehazing Network (DCPI-AODNet) for Single Image Dehazing.-Agentic Broad-Area Search: Evaluating Multimodal Large Language Models for Geospatial Object Identification and Enumeration in Overhead Imagery.-Loop Closure Detection Revisited: A Clustering Perspective.-The Limitations of Image Features from Satellite Imagery Training Datasets.-Synthetic Aperture Radar Change Detection as a Source of Ground-Truth Annotation for Machine Learning Deforestation Detection in the Amazon Using Multispectral Satellite Imagery.-Real-Time 2D Mapping and Navigation on an Indoor Autonomous Vehicle (AV) Platform with RPLIDAR A3 and HectorSLAM.-Towards the Segmentation-Guided Generation of 3D MRA Dataset for Aneurysm Detection.-Hypoxic Ischemic Encephalopathy Diagnosis and Lesion Segmentation Using 3D Heterogenous Ensemble Models.-NeuroGleam: Illuminating Small Vessel Disease Detection Through Deep Learning Based Segmentation of Brain MRI White Matter Hyperintensities.-Multimodal Fusion of Imaging and Multi-Omics Data for Enhanced Breast Cancer Detection.-Wavelet Scattering Features based Colon Cancer Histology Classification.-imgs2imgs: Improving Visual Consistency in Multiview Image Editing.-Quantifying Error Propagation and Recovery in Object Detection for Autonomous Vehicles: a Markovian Approach.-Restorable Segmentation Synthesis Using Fourier Descriptors.-Benchmarking Few-Shot Methods for Rare Target Classification in PlanetScope Imagery.
