computer vision

A PyCon 2012 talk by Kurt Grandis on building an automated water-squirting turret that uses Python and computer vision to detect and drive off squirrels from a bird feeder.

The first lecture of Stanford's CS231n course on convolutional neural networks for visual recognition. It surveys the history and key challenges of computer vision, from the Cambrian explosion and Hubel and Wiesel to ImageNet.

Andrew Ng presents unsupervised feature learning and deep learning, showing how algorithms can learn representations from unlabeled data instead of hand-designed features. He gives examples in vision and audio, including visualizations of features learned by successive layers.

Andrej Karpathy, then Tesla's head of AI, gives a CVPR 2021 workshop keynote on how Autopilot's neural networks work. A large part covers the move to a vision-only stack, including why radar was dropped and how vision produces depth estimates.

The first lecture of Stanford's CS231n Winter 2016 course on convolutional neural networks for visual recognition, introducing the course and the historical context of computer vision.

Andrej Karpathy's CVPR 2021 presentation on Tesla's Autopilot, covering its vision-based approach, sensor choices and the scale of Tesla's deep learning compute cluster. This upload is a lower-quality copy of the talk.

Mike Pound explains the discrete cosine transform and how it underpins JPEG compression. It is part two of Computerphile's JPEG series.

A Computerphile video in which Dr Mike Pound and PhD student Lewis Stuart explain 3D Gaussian splatting, a quick technique that turns photos of a scene into a 3D model without much compute. It includes a demo and compares the approach to NeRFs.

Rob Miles explains on Computerphile how language models generate plausible text and how transformers improved on recurrent networks, with a look at what comes next.

A short 2009 amateur video showing an HP MediaSmart webcam whose face tracking follows a white face but fails to track a black face. The uploader presents it as humorous.

A three-minute demo of ProFORMA, a system from Cambridge that builds textured 3D models live from a video of an object.

A long livestream of George Hotz live-coding a toy SLAM implementation, twitchslam, over two days in May 2018. It is an unedited session of Python programming with computer vision and visualization.

A demo of PREDATOR (TLD), a real-time object tracker that learns the appearance of its target and improves while it tracks. The technology was later developed commercially.

The first episode of the BBC's 1982 television series The Computer Programme, introducing home computing to a general audience. It surveys mainframes, early personal computers such as the BBC Micro and Commodore PET, and the changing computer market of the time.

A University of Tokyo demo of a rock-paper-scissors robot that always wins, using high-speed vision to read the human hand and respond faster than people can perceive.

Short Airbus footage of its ATTOL project, in which a test aircraft performed a fully automatic, vision-based take-off at Toulouse-Blagnac. The crew ran eight take-offs over several hours.

Michael Steil and Christian Hessmann's 34C3 talk explaining the Apollo Guidance Computer end to end: hardware design, operating system, and mission software.

Andrej Karpathy, then Tesla's head of AI, presents at ScaledML 2020 on how Tesla builds full self-driving using vision-based neural nets and fleet data collection. He covers training custom detectors and neural nets gradually replacing traditional code.

Veritasium visits the US Navy's Crane facility to test top night vision goggles. It explains how image intensifier tubes and thermal or digital systems work and compares the generations of the technology.

A showcase of the author's fifth-generation quad-core computer built entirely from Minecraft redstone, after two years of development. It features cache, a hardware stack, a GPU driving a 15x15 display, and serial I/O ports.

FortNine looks at why Tesla Autopilot has rear-ended motorcyclists, speculating that the vision system misreads motorcycle taillights as a distant car. It argues that radar and other sensors would help.

An excerpt from Hal Abelson's opening SICP lecture introducing computer science and contrasting declarative and imperative knowledge. He argues that computer science is not really a science and is more like magic or engineering of ideas.

Luke Miani removes the screen from a MacBook Air and tests using it headless as the host for an Apple Vision Pro virtual display.

The CVPR 2016 oral demo of Face2Face, a system that captures a person's face and reenacts another video's face in real time using only RGB video. The authors frame it as research on modern vision and graphics.

Rob Miles explains "glitch" tokens, odd strings in large language model vocabularies that make the model produce gibberish. He covers how they likely arise from tokenizer training data, such as Reddit usernames, that never appeared in the model's own training.

Ben Eater's short update announcing his plan to build another 8-bit computer from scratch on camera. It kicks off the breadboard computer series that became widely known.

Robert Wills of TNMoC explains the Apollo Guidance Computer hardware and software, then walks through the Apollo 11 landing procedure step by step, including the 1202 alarms.

A 1953 US Navy training film explaining the basic mechanisms in mechanical fire control computers aboard ships. It shows how gears, differentials and cams solved mathematical problems in permanent mechanical form.

A 1953 US Navy training film on the basic mechanisms in fire control computers: shafts, gears, cams and differentials. It shows how these mechanical parts compute ship and target firing solutions.

A full-day recording of the University of Utah computer science department's 50th anniversary event with panels, keynotes and talks by alumni and pioneers. It includes Ivan Sutherland's presentation on single flux quantum electronics.

A SIGGRAPH Asia 2013 research demo of 3-Sweep, a tool for extracting editable 3D objects from a single photo using a few user strokes. It predates modern AI image tools.

CuriousMarc's Apollo Guidance Computer restoration series looks at core rope memory, the ROM that held the Moon landing programs. Mike Stewart builds a custom reader to recover lost AGC programs.

Mobileye co-founder and CTO Amnon Shashua explains what goes into sensing for autonomous driving and whether end-to-end deep learning can work for it. He argues for decomposing the problem.

A short demo from the Ishikawa Komuro Lab of a high-speed robot hand paired with a 1000fps vision system performing dexterous manipulation. It shows what very low-latency vision and actuation can do.

A short demo of the Deep Visualization Toolbox, which shows the activations inside a convolutional neural network as it processes live camera input. Viewers see individual units responding to edges, faces and text.

Marques Brownlee shares his hands-on impressions of Apple's $3500 Vision Pro headset after trying it. He covers the spatial computing experience, eye tracking, and how it compares to other devices.

A 1969 BBC Tomorrow's World segment about Nellie, a school computer used by pupils at Forest Grammar School to solve equations and play music. Students learn programming and maintenance, and act out binary arithmetic.

A short video accompanying the CVPR 2018 paper Learning to See in the Dark. It shows a neural network producing sharp images from very low-light raw sensor data, compared against denoising approaches.

Steve Mould explains motion amplification and video magnification, techniques that find tiny changes in video invisible to the eye and exaggerate them. He covers the MIT research and industrial uses such as diagnosing vibrating machinery.

The video accompanying the SIGGRAPH 2012 paper on Eulerian Video Magnification. It amplifies subtle changes in ordinary video, such as skin color shifts from a heartbeat, to reveal pulse and tiny motions.