Posts Tagged ‘image processing’
Despite constantly increasing safety measures and technological improvements in storage and transfer sites around the world, human error, malfunction or wear of equipment occasionally result in spills of liquid substances such as oils or chemicals.
Each spill has a number of consequences, the extent of which will depend largely on the size of the spill and the site where it occurs. There are possible environmental effects, the cost of cleaning, loss of product and inability to use the site while the cleaning operation is going on. A significant spill may however also have a dramatic effect on the company’s reputation and the value of its shares when it causes negative media coverage.
VanAI has developed an early spill detection system which uses live camera images and a proprietary self-learning algorithm to detect a spill and produce an alarm within seconds. This system has been tested in collaboration with a global tank storage provider. The main capabilities are:
- Detection of spills from hoses not connected to an inlet, or spills from outlets without a hose connected
- Spills caused by wear or improper use of coupling systems
- Detection of liquids spraying from tears or punctures in hoses/pipes
For best accuracy, a dedicated camera with the algorithm running as embedded software can be used. The system is designed to be an out of the box solution which can be easily integrated with an existing surveillance/monitoring infrastructure using standardized protocols.
Key areas of application are sites where bulk liquids are transferred between tanks on land, ships or land vehicles and wherever liquids are used, produced or sold.
- Faster response times than a human operator, reducing the number of medium and large incidents
- Investing in new technology to prevent environmental incidents sends a positive message to clients and the general public
- Rather than being a single-function alarm system, high quality smart cameras can be used as an extension to the existing surveillance infrastructure and can perform additional detection tasks
Please note that the app is in Dutch and that you may or may not be able to find the app in any stores other than the Dutch one. Please feel free to contact us if you have trouble downloading. We can also send you a English app for Android demonstrating only the library’s functionality.
A new version for iOS and Android with improved OCR accuracy and accepting more barcode types will be available soon! The BlackBerry version unfortunately still only features barcode scanning. We’re waiting for the OS to be updated to allow some of the required image overlay features.
For a customer in the agricultural sector, a computer vision system will be developed which observes an organic product on a conveyor belt and produces a qualitative description of the product.
- Preliminary study of the effect of different types of (visible, IR, UV) lighting which helps the detection of small blemishes/deceases.
- 3D object modeling to produce an accurate size/weight estimation
- Texture and color analysis
- Hardware integration, allowing a robotic arm to grab the product from a fast-moving conveyor belt.
- Feedback from a weight sensor in combination with a self-learning algorithm to produce accurate weight estimates.
In cooperation with Innovattic, we are creating a set of mobile application for the Dutch Police which will be part of the Stop Heling project. This project aims to decrease the sales of stolen goods by allowing people to register the serial numbers of their (electronic) belongings. When consumers or police officers come across goods of unknown origin, the apps will make it easy for them to see whether these goods have been reported stolen.
VanAI will be developing the image analysis part of the apps that will allow people to automatically scan serial numbers of their belongings using barcodes or through OCR (text recognition) on the actual serial number. The app needs to be available for the iPhone, Android and BlackBerry, posing several developent challenges.
Advanced vision algorithms allow for a fully automated production inspection, detecting items that are bruised or cracked and selecting premium food items based on surface patterns or shape. Additional sensors can be integrated seamlessly to allow an even more accurate inspection.
- Automatic inspection allows you to make thousands of accurate inspections per minute, easily outperforming a human observer.
- Train a system yourself to new situations or products simply by presenting examples of what you think is “good” and what is “bad”. A self-learning system can continue to learn and adept itself, whereas traditional systems would constantly require expert consultancy.
The above is possible for nearly all types of product; for example an automatic inspection of food products (either with or without packaging), clothers or even complex electronics or PBCs can be inspected with an accuracy equal to or better than that of a manual inspection.
See this page for more information on quality inspection of agricultural producs.
A recent project has demonstrated the effectiveness of AI driven computer vision for agricultural applications.
- Crops (or any kind of organic substance) can be reliably detected, measured and classified/categorized. AI algorithms are surprisingly good at handling problem spaces with fuzzy or ill-defined boundaries. For example, no two potatoes are exactly alike and the quality of a potato is determined by a number of subjective criteria. Nevertheless, smart algorithms could reliably detect, model the 3D shape and classify the quality.
- Pre-harvest analysis is possible for several types of crops. This may allow you to better plan/schedule the harvest and increase your revenue.
- During the entire growth process a large number of variables will determine the final harvest results, many of which can be controlled (humidity, temperature, air flow, fertilizer). Determining the best settings for all these parameters at any point in time is something that’s often done bases solely on experience and “feeling” with the process. AI algorithms can capture this experience and then improve on it by self-learning from results, which in the end inevitably leads to a more optimal system.
- Post-harvest analysis using smart algorithms can improve the sorting of your product; shape, color, texture, bruises, deceases can all be detected as well as a trained worker could do it, in many cases an automatic sorting could be better due to the additional sensor possibilities that are available to a computer (IR, UV, textured light, etc).
Though we currently live in a world with cameras observing us wherever we go, the number of CCTV operators that are monitoring the cameras is so small that hardly any offences are caught through live observations. In the case of a terrorist threat or a physical assault this means the usefulness of the camera network is limited. Automatic detection of suspicious behaviour from camera images is a complex, but more and more feasible task. By only presenting the human operators with camera images showing possible suspicious behaviour, the possibilities of early intervention increase dramatically.
Some example of what AI computer vision algorithms are currently capable of:
- Tracking people / vehicles over multiple camera streams. Unusual paths can be detected and visualized.
- Objects left by people can be detected.
- Group behaviour can be analyzed and classified as alarming or not.
- Detection of specific actions (breaking a window, hitting someone)
As part of a recent project ( more info ) VanAI has created an SDK for barcode reading and text recognition using proprietary algorithms. There are several commercially available solutions with similar functionality, but we believe our implementation offers some strategic advantages:
- Multi-platform support thanks to implementation in C and Java. Runs on nearly anything; Windows, Linux, OS X, iOS (iPhone), Android, BlackBerry, smart cameras (e.g. Axis).
- The architecture was created with mobile platforms in mind and is thus very fast and uses little resources.
- Market leading OCR products are expensive and using their products as part of your own solution requires complex negotiations with large companies.
By using this already developed SDK, we can create custom applications quickly and at low costs, including application for mobile platforms (in cooperation with our partner Innovattic)
You may also use this SDK as part of your own application; all types of licencing models are open for negotiation.
In today’s world of ever increasing availability of video and image content, the need for automatic annotation and interpretation of this data is becoming a serious issue. Technology is available for detection of objects, scenes, movements and styles, making large archives of data accessible once again.
A large video archive has to be made accessible and searchable by scene type. Machine vision technology is used to interpret the content fully automatically and is able to extract the scene settings (indoor, beach, city,..) and objects (people, cars, the White House, logos,…).
- Enormous reduction in labor cost; no manual annotation is needed.
- Easily extendable, a new search category can be added with little effort.
- Better temporal resolution, an object only has to be visible a single frame of a video to be detected.