Vision is a very important part of the human anatomy. It provides us the capability to gauge the environment and detect the threats around us. At the same time, it allows us to evaluate things in our environment and manipulate them.
Most of the technological advancements that are moving us towards Artificial Intelligence attempts to replicate human senses. Machine Vision deals with trying to replicate or even better what the human eye can do. Therefore, Machine Vision is being used to undertake many of the tasks that human use their eyes to do including inspection, evaluation and identification.
Startups are beginning to use these technologies to automate tasks in various industries and opening up potentially larger business avenues.
Agriculture perhaps depends on a most amount of visual inspection. Farmers regularly visually inspect and see if crops are infected, damaged or even to track growth and ripeness. They are also tracking progress through visual inspection.
Agritech startups are trying to automate some of the traditional agricultural processes to reduce human effort. Crop monitoring has always been crucial. Agritech startups are using machine vision in crop monitoring. Crop parameters such as color, health, quality are monitored using these crop monitoring systems. A lot of human effort and time is saved because of the automation. Agricultural drones with high resolution machine vision sensors are used for betterment of plant monitoring systems.
Bangalore based Intello labs is offering its product for identifying the disease of the crop. This crop inspection product identifies the disease from the captured image, identifies the cause of the disease. It also provides recommendations for the cure of the disease.
Another Bangalore based startup Agricx is using machine vision for measuring vital parameters of potatoes. They are using different grading algorithms for determining quality of potatoes.
Crop genetics is also benefited from technology. High resolution machine vision cameras measure plant characteristics. The results are used in developing a database which helps in genetic improvements of crops. This process is known as ‘High Throughput Plant Phenotyping’(HTPP).
You could not possibly drive a vehicle around the town without being able to see. With the advent of cab aggregators and ride sharing services, people are fast realizing how much they hate driving. This is causing a lot of self-driving startups to emerge. While the initial approach was to try sound based mapping, it has become clear that we have built the world – the road, the road signs, etc. for human vision.
Sensing the environment and identifying the objects in the environment is the primary task for a self-driven vehicle. An efficient self-driven vehicle identifies the obstacles and navigates without human help. Machine vision helps in creating obstacle detection systems in self-driven vehicles. Google backed Waymo has even developed a bike recognition algorithm using machine vision. Computer vision techniques used in self-driven vehicles are generally backed by a neural network. Tesla is working on ‘Tesla Vision’ which analyses the surrounding environment for detecting objects in the environment.
Collision detection systems ensure better road safety. They are a part of self-driven as well as normal vehicles. In these collision detection systems real time data is captured and analyzed to avoid the collision. Indian startups such as Accelo Labs and Netradyne are working on IoT devices coupled with machine vision to improve road safety. Accelo is using computer vision for generating a pre collision alert. Development of such innovative solutions would definitely help in reducing road accident deaths in India.
Security and Surveillance has depended on human oversight forever. Teams of people have been hired in the past just to find individuals.
Automation is helping in creating better security systems. Startups are using machine vision to see and understand the environment and activities happening in the environment. Face recognition and detection systems are an inevitable part of these security systems. The automated security systems would be of three types:
Security systems on the personal devices
Security systems installed at public places
Special purpose security systems
Bangalore based startup ‘Uncanny Vision’ is using computer vision and IoT based surveillance cameras for retail outlets and ATMs. This system interprets the video output as information. The video analysis algorithms are used for creating a more efficient security system.
Healthcare is another area where our ability to spot anomalies is critical to proper care. We often use a variety of sensor based imaging techniques to look into the human body. Those images themselves can be analysed through machine vision.
Bangalore based Niramai has developed a software for early detection of cancer in woman using high resolution thermal imaging. Zebra medical vision, an Israel based startup is using proprietary database of imaging scans to create software that analyzes data in real time.
Capsule based endoscopies for the early detection of colon cancers are under research in USA. A small camera is inserted in the capsule which enters the digestive system to have better images of the digestive system. This method of endoscopy is still under research.
Another Israel based startup has developed a device ‘Orcam’ for visually impaired people. The camera in the device reads aloud the text and identifies different objects. This innovative solution is a boon for people with a visual or reading disability.
Fashion industry is widely dependent on the visual appeal of the product. Image intelligence and analytics provide recommendations for online retail platform. Machine vision algorithms are used for scanning the product images which are further classified into different categories using the algorithm. Startups like ‘Stylumia’ are working on fashion intelligence. They claim that source of their intelligence is images, user behavior and textual data on the websites. They use machine vision and text analytics algorithms together to generate insights on their platform.
Retail stores have barcode and object scanners. Machine vision is used for verifying the authenticity of the users. Machine vision is also used as a part of surveillance in retail stores. Object scanners monitor the supply chain activity.
Manufacturing automation is highly dependent on finding objects and processing them. When automation began a huge part of this was undertaken through a variety of sensors. 3D machine vision is being used for monitoring surfaces and coordinates of the manufacturing objects. Machine vision enabled robots are used to perform pick up and place tasks. These robots locate and arrange the parts of assembly.
Companies like Allied Vision are manufacturing embedded cameras which can withstand vibrations, shock and heat. These cameras are high performance oriented and cost effective. Vision sensors coupled with efficient embedded software are important in a manufacturing environment.
These are some of the interesting areas of work that we are witnessing in relationship to Machine Vision. Are you working on something different in this area?