The field of computer vision is rapidly advancing, with a focus on developing more accurate and efficient systems for tasks such as 3D geometric perception, object tracking, and facial micro-expression analysis. Recent research has explored the use of event-based cameras, which offer high temporal resolution and low latency, making them well-suited for applications such as robotics and human-computer interaction. Notable papers in this area include ViPE, which introduces a versatile video processing engine for 3D geometric perception, and DynamicPose, which presents a retraining-free 6D pose tracking framework. Other innovative works include GazeDETR, which proposes a novel end-to-end architecture for gaze detection, and FAMNet, which integrates 2D and 3D features for micro-expression recognition. These advancements have the potential to significantly improve the performance and efficiency of computer vision systems, enabling new applications and use cases.
Advances in Computer Vision and Event-Based Systems
Sources
ViPE: Video Pose Engine for 3D Geometric Perception
DynamicPose: Real-time and Robust 6D Object Pose Tracking for Fast-Moving Cameras and Objects
Exploring Spatial-Temporal Dynamics in Event-based Facial Micro-Expression Analysis
Temporal and Rotational Calibration for Event-Centric Multi-Sensor Systems
HOMI: Ultra-Fast EdgeAI platform for Event Cameras
GazeDETR: Gaze Detection using Disentangled Head and Gaze Representations
Sub-Millisecond Event-Based Eye Tracking on a Resource-Constrained Microcontroller
FAMNet: Integrating 2D and 3D Features for Micro-expression Recognition via Multi-task Learning and Hierarchical Attention
GazeProphet: Software-Only Gaze Prediction for VR Foveated Rendering
OmniSense: Towards Edge-Assisted Online Analytics for 360-Degree Videos
6-DoF Object Tracking with Event-based Optical Flow and Frames
ListenToJESD204B: A Lightweight Open-Source JESD204B IP Core for FPGA-Based Ultrasound Acquisition systems
Scalable FPGA Framework for Real-Time Denoising in High-Throughput Imaging: A DRAM-Optimized Pipeline using High-Level Synthesis
Heatmap Regression without Soft-Argmax for Facial Landmark Detection
\textit{adder-viz}: Real-Time Visualization Software for Transcoding Event Video