Explore the cutting-edge research on large-scale visual inferences in this guest presentation by Dr. Zhuowen Tu from the University of Central Florida. Delve into the fascinating world of bottom-up attention-guided and contextual approaches to visual processing. Learn about unsupervised object discovery, computational neuroscience, and attention points. Discover how multiple instance learning, fixed point theory, and manifold learning contribute to advanced visual recognition systems. Gain insights into gesture recognition, self-diffusion classification, and their applications in modern computer vision. This 58-minute talk covers a wide range of topics, from neuroscience-inspired models to practical implementations in object detection and recognition, providing a comprehensive overview of the latest advancements in visual inference technologies.