Tuesday, November 29, 2016

MIT Creates AI Able to See Two Seconds Into the Future


Massachusetts Institute of Technology announced its new artificial intelligence. Based on a photograph alone, it can predict what’ll happen next, then generate a one-and-a-half second video clip depicting that possible future.

When we see two people meet, we can often predict what happens next: a handshake, a hug, or maybe even a kiss. Our ability to anticipate actions is thanks to intuitions born out of a lifetime of experiences.

Machines, on the other hand, have trouble making use of complex knowledge like that. Computer systems that predict actions would open up new possibilities ranging from robots that can better navigate human environments, to emergency response systems that predict falls, to Google Glass-style headsets that feed you suggestions for what to do in different situations.

This week researchers from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) have made an important new breakthrough in predictive vision, developing an algorithm that can anticipate interactions more accurately than ever before.



Generating Videos with Scene Dynamics

Carl Vondrick
 Hamed Pirsiavash
University of Maryland Baltimore County
 Antonio Torralba

NIPS 2016


We capitalize on large amounts of unlabeled video in order to learn a model of scene dynamics for both video recognition tasks (e.g. action classification) and video generation tasks (e.g. future prediction). We propose a generative adversarial network for video with a spatio-temporal convolutional architecture that untangles the scene's foreground from the background. Experiments suggest this model can generate tiny videos up to a second at full frame rate better than simple baselines, and we show its utility at predicting plausible futures of static images. Moreover, experiments and visualizations show the model internally learns useful features for recognizing actions with minimal supervision, suggesting scene dynamics are a promising signal for representation learning. We believe generative video models can impact many applications in video understanding and simulation.

Monday, November 28, 2016

Build hardware synchronized 360 VR camera with YI 4K action cameras



YI 4K Action Camera is your perfect pick for building a VR camera. The camera boasts high resolution image detail powered by amazing video capturing and encoding capabilities, long battery life and camera geometry. This is what makes us stand out and how we are recognized and chosen as a partner by Google for its next version VR Camera, Google Jump - www.yijump.com
There are a number of ways to build a VR camera with YI 4K Action Cameras. The difference being mainly how you control multiple cameras to start and stop recordings. In general, we would like all cameras to start and stop recording synchronously so you can easily record and stitch your virtual reality video.
The easiest solution is to manually control the cameras one-by-one. It is convenient and quick however it doesn’t guarantee synchronized recording.
A better solution therefore is to make good use of Wi-Fi where all cameras are set to work in Wi-Fi mode and are connected to a smartphone hotspot or a Wi-Fi router. Once setup is done, you should be able to control all cameras with smartphone app through Wi-Fi. For details, please check out https://github.com/YITechnology/YIOpenAPI
Please note that this solution also comes with its limitations. For instance, when there are way too many cameras or Wi-Fi interference happens to be serious, controlling the cameras via smartphone app can sometimes fail. Also, synchronized video capturing is not guaranteed since Wi-Fi is not a real-time communication protocol.
You can also control all cameras with a Bluetooth-connected remote control. The limitations however are similar to that with Wi-F solution.
There are also solutions which try to synchronize video files offline after recording is finished. It is normally done by detecting the same audio signal or video motion in the video files and aligning them. Since this kind of the solutions do not control the recording start time, its synchronization error is at least 1 frame.


In this article, we will introduce a solution to solve synchronization problem using hardware. We do this by connecting all cameras using Multi Endpoint cable where recording start and stop commands are transmitted in real-time among the cameras, in turn, creating a high-resolution and synchronized virtual reality video.

Saturday, November 26, 2016

Breathing Life into Shape (SIGGRAPH 2014)

This has serious implications in security, interrogation, marketing, health and medical.

Soon high res 3D Depth cameras will be cheap and common place.

Pinlight Displays: Wide Field of View Augmented Reality Eyeglasses (SIGG...

Slim near eye display using pinhole aperture arrays

Boom, Mike Drop.

This just got trivial.

Friday, November 25, 2016




Spoiler Alert, Basically they use the Raspberry Pi to connect to the CCTV IP camera over RTSP and then send a live stream up to youtube. Up, out though your firewall and NAT to youtube or any service that will accept RTMP.

Most of this is a good NOVICE GUIDE to setting up and configuring the Pi and buying a IP camera from them.

  1. You can download the source code for the BASH script here.



COMMAND="sudo ffmpeg -f lavfi -i anullsrc -rtsp_transport tcp -i ${RTSP_URL} -tune zerolatency -vcodec libx264 -t 12:00:00 -pix_fmt + -c:v copy -c:a aac -strict experimental -f flv ${YOUTUBE_URL}/${YOUTUBE_KEY}"

if sudo /usr/bin/pgrep $SERVICE > /dev/null
        echo "${SERVICE} is already running."
        echo "${SERVICE} is NOT running! Starting now..."

They have a bunch of neat Pi Video projects on their web site.

Wednesday, November 23, 2016

Chronos 1.4 high-speed camera up to 21,600fps

Chronos 1.4 is a purpose-designed, professional high-speed camera in the palm of your hand. With a 1.4 gigapixel-per-second throughput, you can capture stunning high-speed video at up to 1280x1024 resolution. Frame rate ranges from 1,057fps at full resolution, up to 21,600fps at minimum resolution.

Features and specs

See the full specs in the Chronos 1.4 Datasheet
  • 1280x1024 1057fps CMOS image sensor with 1.4Gpx/s throughput
  • Higher frame rates at lower resolution (see table below)
  • Sensor dimensions 8.45 x 6.76mm, 6.6um pixel pitch
  • Global shutter - no “jello” effect during high-motion scenes
  • Electronic shutter from 1/fps down to 2us (1/500,000 s)
  • CS and C mount lens support
  • Focus peaking (focus assist) and zebra exposure indicator
  • ISO 320-5120 (Color), 740-11840 (Monochrome) sensitivity
  • 5" 800x480 touchscreen (multitouch, capacitive)
  • Machined aluminum case
  • Record time 4s (8GB) or 8s (16GB)
  • Continuous operation on AC adapter (17-22V 40W)
  • 1.75h runtime on user-replaceable EN-EL4a battery
  • Gigabit ethernet remote control and video download*
  • Audio IO and internal microphone*
  • HDMI video output*
  • Two channel 1Msa/s waveform capture*
  • Storage: SD card, two USB host ports (flash drives/hard drives), eSATA 3G
  • Trigger: TTL, switch closure, image change*, sound*, accelerometer*
  • Low-noise variable-speed fan - camera can run indefinitely without overheating


Monday, November 21, 2016

Sunday, November 20, 2016

ELP contact info

Maker of USB webcam boards.