State-of-the-Art Deep Learning-Based Object Detection in 2D

Explore the cutting-edge advancements in deep learning-based object detection for 2D images in this 43-minute conference talk by Dr. James G. Shanahan from Church and Duncan Group and UC Berkeley. Dive into the evolution of object detection techniques, from two-stage approaches to more efficient one-stage models, and learn how mean average precision has improved dramatically over the past five years. Examine state-of-the-art object detection systems, including two-stage, one-stage, and transformer-based approaches. Gain practical insights on implementing and upgrading detection pipelines, with examples provided in Jupyter Notebooks using PyTorch, Python, and OpenCV. Discover the exciting applications of object detection in fields such as self-driving cars, smart cities, and cancer detection. While focusing primarily on 2D digital images from cameras and videos, the talk also introduces object detection in 3D point clouds, providing a comprehensive overview of this rapidly advancing field in computer vision.