New artificial intelligence system mimics how humans visualize and identify objects

The approach is made up of three broad steps. First, the system breaks up an image into small chunks, which the researchers call “viewlets.” Second, the computer learns how these viewlets fit together to form the object in question. And finally, it looks at what other objects are in the surrounding area, and whether or not information about those objects is relevant to describing and identifying the primary object.

Dec 18, 2018 - By UCLA Samueli Newsroom

Researchers from UCLA Samueli School of Engineering and Stanford have demonstrated a computer system that can discover and identify the real-world objects it “sees” based on the same method of visual learning that humans use.

The system is an advance in a type of technology called “computer vision,” which enables computers to read and identify visual images. It is an important step toward general artificial intelligence systems–computers that learn on their own, are intuitive, make decisions based on reasoning and interact with humans in a more human-like way. Although current AI computer vision systems are increasingly powerful and capable, they are task-specific, meaning their ability to identify what they see is limited by how much they have been trained and programmed by humans.

Even today’s best computer vision systems cannot create a full picture of an object after seeing only certain parts of it–and the systems can be fooled by viewing the object in an unfamiliar setting. Engineers are aiming to make computer systems with those abilities–just like humans can understand that they are looking at a dog, even if the animal is hiding behind a chair and only the paws and tail are visible. Humans, of course, can also easily intuit where the dog’s head and the rest of its body are, but that ability still eludes most artificial intelligence systems.

Current computer vision systems are not designed to learn on their own. They must be trained on exactly what to learn, usually by reviewing thousands of images in which the objects they are trying to identify are labeled for them.

Computers, of course, also cannot explain their rationale for determining what the object in a photo represents: AI-based systems do not build an internal picture or a common-sense model of learned objects the way humans do.

The engineers’ new method, described in the Proceedings of the National Academy of Sciences, shows a way around these shortcomings.

The approach is made up of three broad steps. First, the system breaks up an image into small chunks, which the researchers call “viewlets.” Second, the computer learns how these viewlets fit together to form the object in question. And finally, it looks at what other objects are in the surrounding area, and whether or not information about those objects is relevant to describing and identifying the primary object.

To help the new system “learn” more like humans, the engineers decided to immerse it in an internet replica of the environment humans live in.

“Fortunately, the internet provides two things that help a brain-inspired computer vision system learn the same way humans do,” said Vwani Roychowdhury, a UCLA professor of electrical and computer engineering and the study’s principal investigator. “One is a wealth of images and videos that depict the same types of objects. The second is that these objects are shown from many perspectives–obscured, bird’s eye, up-close–and they are placed in different kinds of environments.”

Photo: The “computer vision” system can identify objects based on only partial glimpses, like by using these photo snippets of a motorcycle. (credits: UCLA Samueli School of Engineering / Stanford)

To develop the framework, the researchers drew insights from cognitive psychology and neuroscience.

“Starting as infants, we learn what something is because we see many examples of it, in many contexts,” Roychowdhury said. “That contextual learning is a key feature of our brains, and it helps us build robust models of objects that are part of an integrated worldview where everything is functionally connected.”

The researchers tested the system with about 9,000 images, each showing people and other objects. The platform was able to build a detailed model of the human body without external guidance and without the images being labeled.

The engineers ran similar tests using images of motorcycles, cars and airplanes. In all cases, their system performed better or at least as well as traditional computer vision systems that have been developed with many years of training.

The study’s co-senior author is Thomas Kailath, a professor emeritus of electrical engineering at Stanford who was Roychowdhury’s doctoral advisor in the 1980s. Other authors are former UCLA doctoral students Lichao Chen (now a research engineer at Google) and Sudhir Singh (who founded a company that builds robotic teaching companions for children).

Singh, Roychowdhury and Kailath previously worked together to develop one of the first automated visual search engines for fashion, the now-shuttered StileEye, which gave rise to some of the basic ideas behind the new research.

Read more

Comments

No comments to display.

Related posts

Toyota Mobility Foundation announces finalists in $4 million global Mobility Unlimited Challenge

The Challenge invited engineers, innovators, and designers from across the world to submit designs for game-changing technologies, incorporating intelligent systems, to improve the mobility and independence of people with lower-limb paralysis. Central to the Challenge is the importance of collaboration with end-users to develop devices which will integrate seamlessly into users’ lives and environments, while being comfortable and easy to use, enabling greater independence and increased participation in daily life.

PRIMA Call for Proposals: Bridging the gap between potential and actual irrigation performance in the Mediterranean

In the Mediterranean basin, water balances and water used for irrigation including irrigation performance is often lower than potential at the farm and at the water users association level. This gap is evident for average performance, but it is particularly important when the dispersion among farmers and their associations is considered. Attaining Sustainable Mediterranean irrigation requires to assess the real water balances in the area of irrigation, amount of water available for irrigation ensuring that water levels are environmentally safe for ecosystems and other needs. Furthermore, optimising irrigation performance requires to combine technological, managerial and governance innovations taking into account the availability, safety and quality of water resources, existing institutional arrangements, as well as crop agronomy planning.
Application Deadline in a month

PRIMA Call for Proposals: Agro-food Value Chain 2019

Proposals should focus on products with high added value, most exposed to fraud and valorise productions that have been discontinued but hold nutritional value and are important for biodiversity conservation. Proposal should also aim at creating trans-national networks of laboratories, validating and harmonizing protocols, sharing open-data on food products of the Mediterranean tradition, providing a robust support for certification and control bodies, and building capacity in key enabling technologies related to authenticity and traceability.
Application Deadline in a month

African Development Bank Group approves $25 million Equity Investment in Fund for Renewable Energy Projects in Africa

ARPF projects will focus on mature technologies including wind, solar PV, small to medium hydro, geothermal and biomass. These would include grid-connected independent power producers (“IPPs”), and decentralized energy projects (commercial & industrial solar, mini-grids and solar home systems companies). The Fund’s strategy is to prioritize projects with a clear timeline to financial close, with emphasis on de-risking early stage greenfield projects.

European call for proposals: Innovative utilisation of aquatic biomass and Novel aquaculture production systems

The proposals should seek to be complementary with or adding to other projects funded under FP7 and Horizon 2020, ERA-NETs, BBI JU or other funding schemes, both ongoing and recently finished. They should reflect awareness of the objectives of running projects in relevant fields to avoid overlap. Proposed projects must demonstrate industry involvement.
Application Deadline in 2 months

European call for proposals: Sustainable and Resilient agriculture for food and non-food systems

Based on the two previous calls, the FACCE SURPLUS partners decided to further advance their successful funding scheme with a third call of the initiative in order to complete the portfolio of projects and filling remaining research and funding gaps. Having in mind the research already conducted in FACCE SURPLUS projects, this call at the end of the initiative aims at further improving collaboration in the area of food and nonfood biomass production and transformation systems.
Application Deadline in 2 months

A Quick Look At The Latest Happenings In The Cloud Integration Market

As an emerging trend of technology the field of cloud Integration is steadily expanding. The global cloud Integration Market is growing approximately at a CAGR of 8% during 2018-2023 and it is expected to grow at a highest CAGR during the forecast period.

A Close Look At The Latest Research Trends Within The 3D Printing Metals Market

3D printing involves the creation of a 3 dimensional object by adding successive layers of a particular material one on top of each other until the overall object has been formed. Metals constitute one of the main materials used for 3D printing.

Call for Applications: Go Green in the City 2019

The Go Green in the City challenge is now entering its 9th year running and our 2019 edition is going to be even bigger, better and bolder! We are looking for students like you from around the world who are willing to put their skills and ideas to the test.
Application Deadline in 4 months