Friday, September 28, 2018

This AI Learned To See In The Dark

The paper "Learning to See in the Dark" and its source code is available here:

Thermal polarimetric imaging.

Researchers demonstrate an example of human identification using conventional and polarimetric thermal cameras. The thermal polarimetric image allows for fine facial details to emerge, researchers said. (U.S. Army Photo)

"Researchers have known for about 30 years that man-made objects emit thermal radiation that is partially polarized, for example, trucks, aircraft, buildings, vehicles, etc., and that natural objects like grass, soil, trees and bushes tend to emit thermal radiation that exhibits very little polarization," Gurton said. "We have been developing, with the help of the private sector, a special type of thermal camera that can record imagery that is based solely on the polarization state of the light rather than the intensity. This additional polarimetric information will allow Soldiers to see hidden objects that were previously not visible when using conventional thermal cameras."

"Prior to our research at ARL, the only way to view humans at night was to use conventional thermal imaging," Gurton said. "Unfortunately, such imagery is plagued by a "ghosting" effect in which detailed facial features required for human identification are lost. However, when polarization information is included in the thermal image, i.e., a thermal polarimetric image, fine facial details emerge, which allows facial recognition algorithms to be applied."

Monday, September 17, 2018

Fwd: OpenMV News

---------- Forwarded message ----------
From: "OpenMV" <>
Date: Sep 17, 2018 8:41 AM
Subject: OpenMV News
To: "John" <>

OpenMV Home -
View this email in your browser

OpenMV Cam H7 Kickstarter Launched


Hi everyone,

The wait is over! The OpenMV Cam H7 Kickstarter has launched!

OpenMV Cam H7 Kickstarter

Help OpenMV make the OpenMV Cam H7 a reality by backing us on Kickstarter!

Copyright © 2018 OpenMV, LLC, All rights reserved.
You are receiving this email because you provided your email address to OpenMV at some point in time. Please opt out if you do not want to receive communications from OpenMV.

Our mailing address is:
PO Box 1577 #22900
Atlanta, GA 30301

Add us to your address book

Want to change how you receive these emails?
You can update your preferences or unsubscribe from this list

Email Marketing Powered by MailChimp

Saturday, August 11, 2018

Machine vision camera

About JeVois

Open-source machine vision finally ready for prime-time in all your projects!
JeVois = video sensor + quad-core CPU + USB video + serial port, all in a tiny, self-contained package (28 cc or 1.7 cubic inches, 17 grams or 0.6 oz). Insert a microSD card loaded with the provided open-source computer vision algorithms (including OpenCV 3.4.2, TensorFlow, Caffe, Darknet, and many others), connect to your desktop, laptop, and/or Arduino, and give your projects the sense of sight immediately.

Monday, July 30, 2018


Television Newspaper & Intramural Television - Scoops magazine UK (1934/1935)

Television Newspaper:  Can it be Done? syndicated comic, Scoops magazine UK (1934/1935)

Sunday, July 29, 2018

My Notes on TV from 2011, Evolution of Television. Future TV.

The Price of a Television has remained relatively constant

Even High End Televisions are oddly unchanged over a 70 year span!

We have added new capabilities, 3D, 4K, vast improvements in color and image quality

But where are we going ?

Of all the screens we have, we can tolerate much longer viewing times on a TV than any other device.

Below are some thoughts on the Content Ecosystem.

John Sokol shared “Lambert Castle Museum, Passaic County Historical Society Library, Video Related items” with you

Lambert Castle Museum, Passaic County Historical Society Library, Video Related items
View album

Monday, April 09, 2018

51 percent of IP cameras are infected with Internet of Things botnet malware

An analysis of roughly 4,400 IP cameras in the U.S. using custom http servers found that just over 51 percent of them are infected by one of four Internet of Things botnet malware families, according to new research.
The majority of these 3,675 compromised cameras, or approximately 64.1 percent, were infected by the IoT botnet Persirai, Trend Micro reported in a blog post on Thursday. Discovered earlier this year, Persirai relies on exploited vulnerabilities to steal credentials and attack other devices.
The remaining affected cameras were found infected by the IoT botnets Mirai (about 27.7 percent), DvrHelper (about 6.8 percent), and TheMoon (about 1.4 percent), the blog post continues. Trend Micro used the Shodan search engine as well as its own research to amass its study sample, though it is not currently clear how recently this analysis took place. (SC Media has contacted Trend Micro for an answer.)

Mirai appears to be spreading fast. A security researcher put online six virtual machines designed to look like ADSL routers running Linux operating systems just like the ones targeted by Mirai—in other words, a set of honeypots.
It took only an average of 15 minutes for these to get hit with Mirai malware, the researcher, who asked to be referred to as "Jack B." to protect his real identity, told me in an online chat. (If you didn't just say "holy shit," you probably should have.)

Monday, February 26, 2018

Nicolas Cage as Everyone - Nick Cage DeepFakes Movie Compilation

How to make deep fakes?

FakeApp can easily create robust, diverse datasets of thousands of images in minutes from both image sets and videos.

Get Datasets
Train AIs
Create Faceswaps
FakeApp makes it easy to observe the progress of trained AIs in real time by posting frequent loss values and training previews.

FakeApp reduces the task of converting the face in a video to a single button process by automatically splitting, converting, and stitching video frames.

FakeApp Makes Faceswaps Easy
FakeApp was designed to make the process of creating realistic faceswaps with deep learning as smooth, simple, and quick as possible. It supports all three steps of the basic faceswap workflow—creating datasets, training AI, and converting videos. Here is a brief video tutorial made by a FakeApp user and some screenshots.

Google built a tiny Street View car to map out one of the world's largest model cities

This is just too cool.

The "Miniatur Wunderland" exhibition, located in Hamburg, Germany, is the world's largest model railway.
To bring new perspective to the massive model railway, Google last year built a miniature version of its Street View car to capture footage within the Miniatur Wunderland with an array of tiny mounted cameras. You can actually see all the various worlds within the Miniatur Wunderland on Google Street View.

HD Analog video technology comparison - HD-SDI, AHD, HD-CVI and HD-TVI

HD Analog technology offers a compelling alternative to traditional analog and IP surveillance options. HD Analog video security solutions deliver megapixel resolution video over coaxial cabling (HD over coax), and are far less complex to install and operate compared to IP.

Since its introduction, several HD Analog formats have emerged, including: HD-SDI, AHD, HD-CVI and HD-TVI. While all HD Analog options offer the same basic benefits compared to analog or IP, each format features a unique set of capabilities and compatibilities.

At this time, there is no “standard” format – or established “winner” – in the HD Analog space. But, it is clear that certain formats are better positioned to succeed – offering clear competitive advantages that translate to significant and measurable benefits for the end user.

How HD Analog Formats Compare
Max Resolution (Pixels)1080p
1920 x 1080 px
1920 x 1080 px
1920 x 1080 px
1920 x 1080 px
Video Output Quality
(What you see)
Very good in daylight
Images not as crisp in low light conditions
Images not
as crisp
Colors not as accurate
Very Good
Sharp images up to 1080p
True colors
Very Good
Sharp images up to 1080p
True colors
Coaxial Cable
1600’ (720p)
1300’ (1080p)
Twisted Pair
(digital converter
DVR Input
AnalogAll analog cameras, limited channels, limited configurationsAll analog cameras, limited channels, limited configurationsLimited models, limited channels, limited configurationsAll analog cameras, on any channel, in any configuration
HD AnalogLike-branded HD-CVI cameras, on any channel, in any configurationLike-branded AHD cameras, on any channel, in any configurationLike-branded HD-SDI cameras, on any channel, in any configurationAny HD-TVI camera, on any channel, in any configuration
IPSupported IP cameras, on select channels, in select configurationsN/AN/ASupported IP cameras, on 2 channels, in any configuration
DVR Hybrid CapabilitiesLimited
Yes, unlimited configurability
Format CompatibilityProprietary technology, only produced by 1 manufacturer, Dahua - out of ChinaOpen
standard, limited manufacturers
standard, multiple manufacturers - are currently exiting the market as alternative HD formats outpace SDI
Open standard, over 100 manufacturer's have launched or are developing solutions using TVI technology


HD-SDI is a digital format over coax usually. It's used in professional Television studios and tends to be very expensive.


HD-TVI stands for High Definition Transport Video Interface.
The TVI standard was developed by a company named Techpoint and released in 2014 and adopted by many manufacturers, most notably Hikvision – Dahua’s primary competitor. Hikvision is the largest manufacturer of HD-TVI in the world. LTS Security is the biggest reseller of their products.


High Definition Composite Video Interface (HDCVI) was developed by Dahua in 2012 and began to saw wide-release in 2014.

The PTZ cameras of the HDCVI type are capable of achieving that RS485 data connection without the use of an extra cable. It's all done right through the same coax cable that's transmitting video.

HD-CVI technology can at times be found masked by behind other names like HD-AVS or HD-MPX.


AHD (Analog High Definition) developed by NextChip, a Korean manufacturer. AHD is interesting because most of its distribution has come in the form of no name brands. However, it was recently adopted by Samsung, making the technology considerably more legitimate as far as mainstream markets and long-term sustainability go. HD-AHD cameras are very cheap and are intended to occupy the low-end market.

Tuesday, January 23, 2018

Facebook Oculus invented a new time unit called the ‘flick’

A flick (frame-tick) is a very small unit of time. It is 1/705600000 of a second, exactly.
1 flick = 1/705600000 second

Here’s a list of numbers into which 1/706,600,000 divides evenly: 8, 16, 22.05, 24, 25, 30, 32, 44.1, 48, 50, 60, 90, 100, 120.

When working creating visual effects for film, television, and other media, it is common to run simulations or other time-integrating processes which subdivide a single frame of time into a fixed, integer number of subdivisions. It is handy to be able to accumulate these subdivisions to create exact 1-frame and 1-second intervals, for a variety of reasons.
Knowing that you should never, ever use floating point representations for accumulated, simulated time (lest your temporal accuracy degrade over time), the std::chrono time tools in C++ are ideal. However, the highest usable resolution, nanoseconds, doesn't evenly divide common film & media framerates. This was the genesis of this unit.

Wednesday, January 17, 2018

Is ATSC 3.0 really a Trojan Horse for next-generation AV and Collaboration?

Is ATSC 3.0 really a Trojan Horse for next-generation AV and Collaboration?

You may have seen the latest headline coming out of CES in Las Vegas last week about the Advanced Television Systems Committee (ATSC) release of ATSC 3.0, or more accurately stated the standards suite of protocols. How does this release impact the Audiovisual and Integrated Experience Association (AVIXA) members and the design of AV systems in general?