Intel演示了在移动平台上硬件加速计算机视觉算法。来源
This research from Intel Labs, St. Petersburg explores the automated design of reconfigurable accelerators based on tools-aided application analysis targeting computationally-intensive media workloads such as the SURF object recognition algorithm.
Google准备发布goggles Chrome扩展。来源
We would like to have some browser extensions so that the user can drag a Web image and drop it in an input box on the toolbar…The search results of the image will be shown
Fraunhofer正在研发一个系统，可以用手机来看周围的电器消耗了多少能源。来源
The technology behind this feature is complex: A server stores pictures of the individual devices, taken from a number of directions. When the function is activated, the cell phone sends the picture taken to the server, which then compares the picture with the ones in its database. As soon as it has recognized the device, it determines the power consumption at the time as reported by the associated power plogg, and sends this information back to the cell phone.
无技术纯转帖
神秘的公司神秘的项目
Many of you have been wondering what exactly we have been working on at Neogence Enterprises and when we are going to launch something. We are planning on a full global launch in October, but we are going to release an open beta around June or July. Even more fun, we are going to release a very early alpha version in a few months. We have an aggressive plan for regular updates and iterations from alpha to full launch, and we want to involve the community. Grow with us.
Head on over to twitter and follow @mirascape if you want to be on the list for early access and some early adopter specials. We aren’t planning on advertising this early alpha access beyond my blog here and a little bit of twitter. I’ll be interested in seeing how many people follow @mirascape (I am fascinated by social networks and the organics of viral word of mouth). We will start doing some announcements and updates on that twitter account once things are ready to open up access. Don’t expect frequent updates though, we are neck deep in development right now, and that has priority.
So, what is the Mirascape™?
Mira: Look! (Spanish), Wonderful or Astonishing (Latin), World (Russian), and Prosperous (Hindi).
The Mirascape™ is the world’s first global augmented reality network, platform, and community. Mirascape™ is about connecting people, places, and things in a way never done before. It is about the convergence of social, local, mobile, and virtual. It is about empowering you, the user, to create and interact with media and the world around you in new ways. It is about engagement and experience.
来源
http://www.neogence.com/
http://mirascape.com/
modiface是一家提供人像美化技术的公司，由多伦多大学的教授Parham Aarabi在2006年创建。modiface目前为20多家公司提供人像美化相关技术。
按照官网的说明，modiface提供人脸检测，五官定位，头发轮廓检测，发色变换，发型变化，face anit-aging，lifting alterations（这两个不会翻译，高手指点一下？），瘦身模拟和超过20种以上的人像美化，上妆效果。所有这一切都可以在几秒内完成，几乎不需要人工干预。
该公司的最大特点我觉得不是他们的技术，而是他们的运作方式。modiface不直接提供面向用户的服务，而是向其他公司提供核心技术。在这个计算机视觉这个仍旧很朝阳的领域里，人人都想着怎么发掘应用，这家公司有点意思，我不做应用了，我只做技术，你们爱怎么用怎么用吧。
阅读全文…
设计GUI很痛苦，编写GUI代码更痛苦，修改GUI代码更更痛苦，测试GUI代码更更更痛苦，所以就有了这么个东西sikuli（死吧！苦力？？），用计算机视觉进行自动GUI测试，以代替无比繁复的点击点击点击点击。
简单的说就是，测试人员编写测试脚本，计算机会自己去检查那个按钮，窗口，什么的到底出现没有。论文里有详细的解释，来自MIT CSAIL。
看起来是个不错的东西，不过作者也提到了，还有一些问题不能解决，比如现在sikuli虽然能检测某个按钮出现没有，但是如果其他地方莫名其妙出现个一个别的什么什么东西，sikuli是无能为力的，因为sikuli只知道到脚本指定的地方去检查，它还没有全屏“自省”的能力。
这里是项目主页。
媒体报道
穿上这坨T恤你就可以站在摄像头前面跟自己玩石头剪子布了。。。看这里，需要翻墙才能看到视频。
阅读全文…
snaptell是一票印度人创建的美国公司。主要业务是手机图像检索。意思就是通过手机摄像头抓拍到的图像搜索相关信息（看名字知道啦，snap->tell）。
官方介绍他们的算法非常之精准，可以对付遮挡，光照不均，扭曲，透视，缩放等等，总之很牛。他们的算法名字叫”Accumulated Signed Gradient” (ASG)，但是我在google scholar和google patent上面都没有搜到，保密工作做得好啊。
官网介绍他们的算法在一次比赛中获得了百分百的识别率。下面有些样例。
阅读全文…
nvidia的CEO黄仁勋(搞不清楚他的中文名字了，貌似是这么念得），在最近的一个访谈里说道：GPU会使计算机视觉变得简单许多。这不是废话么。。。
“The computer would simply know that it is me sitting in front of it. I wouldn’t need to type in a password. It would know it’s me based on what it sees of me, the way I am acting and talking.
“These kinds of capabilities are certainly within the next generation, because we have created a processor for the GPU that makes it possible to do parallel processing so much faster on a PC. We think this GPU technology is going to transform computing in a way that will bring back the magic to consumers.”
上面的引言也是废话，请无视。。。我引用只是为了显得我很专业。。。
这次更新都是废话。
来源
另外推荐一下nvidia的官方博客。值得一看。
Hella Aglaia用Impulse Accelerated Technologies的Impluse C开了一套基于FPGA的人流量检测系统。Impulse的VP声称使用它们的Impulse C，开发速度要比通常使用VHDL或者verilog快上50%，比HDL快80%。
这会不会是一个趋势呢？从理论到算法，从算法到C语言，从C语言到FPGA，从FPGA再到ASIC？？
厂商链接：Impluse，Hella Aglaia
用人脸检测训练时候挑出的非人脸拼成的像素画–cvchina。
最新评论