Video Object Detection At Scale with Reza Zadeh - TWiML Talk #34
Analysis
This article summarizes a podcast episode featuring Reza Zadeh, a Stanford professor and CEO of Matroid. The discussion centers on scaling deep learning, particularly for video object detection. The conversation covers challenges and approaches to scaling deep learning in general and within Matroid's video object detection service. It also touches upon advancements in computer vision, including the use of CPUs, GPUs, and the emerging role of TPUs, as well as a deeper dive into Apache Spark. The article highlights the practical applications of deep learning and the evolution of hardware in the field.
Key Takeaways
“Our conversation focused on some of the challenges and approaches to scaling deep learning, both in general and in the context of his company’s video object detection service.”