Karhunen-Loève Transform (KLT) Algorithm for Tracking - Lecture 10
Explore tracking techniques, including KLT algorithm, for computer vision applications. Learn about translation models, Jacobian matrices, and multi-camera setups.
Scale-Invariant Feature Transform (SIFT) - Lecture 6
Explore SIFT algorithm's principles, advantages, and applications in computer vision, including scale-space representation and feature detection techniques.
Visual Analysis of Extremely Dense Crowded Scenes
Analyzing dense crowd scenes using computer vision techniques for counting, detecting, and tracking individuals. Explores multiple data sources, context, and neighborhood motion to address challenges in extremely crowded environments.
Human Detection, Tracking and Segmentation in Surveillance Video
Explore advanced techniques for human detection, tracking, and segmentation in surveillance videos, focusing on scene-specific learning and occlusion handling for improved accuracy and efficiency.
Action Recognition in Temporarily Untrimmed Videos
Action recognition in untrimmed videos: challenges, datasets, and annotation techniques for improving accuracy in real-world scenarios with ambiguous content and varying video qualities.
Statistical and Spatial Consensus Collection for Detector Adaption
Explore detector adaptation techniques for improved pedestrian detection across domains, focusing on statistical and spatial consensus methods to enhance accuracy and generalization.
Visual Geo-Localization and Location-Aware Image Understanding
Explore visual geo-localization techniques, including image matching, geospatial pruning, and segment localization. Learn about business recognition and multi-hypotheses character recognition for location-aware image understanding.
CAP6412 - Final Project Presentations - Lecture 27
Final project presentations showcasing innovative approaches in video action recognition, offline reinforcement learning, and decision transformer encoders, with discussions on methodologies, results, and implications.
CAP6412 - Advanced Computer Vision - Lecture 24: Project Presentations
Students present their projects, showcasing innovative approaches in computer vision and AI. Topics include object detection, video analysis, and visual question answering, highlighting challenges and novel solutions.
Computer Vision: Tokens and Cross-View Fusion - Lecture 22
Explore tokens, cross-view fusion, and attention in computer vision, examining experimental results and innovations in multi-view learning techniques.
Computer Vision Architecture Evolution: ConvNets to Transformers - Lecture 21
Explore the evolution of vision architectures, comparing SWIN and CNNs, and learn about modernizing ConvNets with macro and micro design changes for improved performance.
Rethinking and Improving Relative Position Encoding for Vision Transformer - Lecture 23
Explore vision transformer improvements through relative position encoding, examining bias modes, piecewise indexing, and 2D calculations for enhanced performance in image processing tasks.
Emerging Properties in Self-Supervised Vision Transformers - Lecture 25
Explore emerging properties in self-supervised vision transformers, covering related topics, model overview, and implementation details for advanced computer vision applications.
Contextual Image Understanding and Scene Identification in Videos - Thesis Defense
Explore image understanding using context, focusing on scene identification and object detection in videos through innovative techniques like Top-Down Visual Tree representation and Random Walk incorporation.
Learning to Navigate in Cities Without a Map
Explore reinforcement learning techniques for autonomous navigation in unmapped urban environments, focusing on value functions, courier tasks, and multi-site experiments.
Computer Vision - Adversarial Attacks and Deepfool Algorithm - Lecture 1
Explore DeepFool algorithm for generating adversarial examples in deep neural networks, covering linearization, multiclass classification, and experimental results.
Instance Segmentation - Lecture 26
Explore instance segmentation in computer vision, covering Mask R-CNN, ROI pooling, and alignment techniques for precise object detection and pixel-level classification.
Exploring Relationships Between Ground and Aerial Views by Synthesis and Matching
Explore relationships between ground and aerial views through synthesis and matching techniques, focusing on architecture, evaluation methods, and practical applications in image retrieval and GPS data.
Image Segmentation Basics and Techniques - Lecture 23
Explore image segmentation basics, including binarization, thresholding, and region-based techniques. Learn algorithms, visualization methods, and practical applications in computer vision.
Cross-View Geo-Localization - Ground-to-Aerial Image Matching
Explore cross-view geo-localization techniques, focusing on ground-to-aerial image matching. Learn about approaches, limitations, and advancements in this field of computer vision.
Object Detection in Computer Vision - Lecture 22
Explore advanced object detection techniques in computer vision, including R-CNN, Fast R-CNN, and Faster R-CNN. Learn about training processes, region proposals, and improvements in detection algorithms.