Meet Risab Biswas, a developer who builds a computer vision model that can detect pathological disease in plants
Risab Biswas's product helps farmers to improve their crop harvests
At a time when millions do not have enough food, nearly 20% of all fruit and vegetable supply is lost during production, according to the Food and Agriculture organisation. To help farmers improve their crop harvests caused by pathogen attacks, Risab Biswas, a Jalpaiguri, West Bengal-based developer, build a computer vision model that can detect pathological disease in plants.
Biswas first gathered plant data from Google images, then used TensorFlow (widely-used machine learning framework in the deep learning space) and Open Vino (Intel’s neural network optimisation toolkit) to build an AI model. Once the images and videos of plants were captured the model is used to identify the cause of the disease, possible cures and preventive measures. To run these solutions, Biswas used Intel 7th Gen i5 NUC mini PC.
Biswas isn’t the only developer to use emerging tech in a bid to save the planet. Intel claims its AI solutions are being used by tens of thousands of developers across the world to make planet Earth a better place for its inhabitants. Another independent developer, Rosemarie Day, has built system using Tensor Flow and Intel’ Neural Compute Stick (NCS) to help environmentalists understand the impact of deforestation on plants. Between 1990 and 2016, planet earth lost 1.3 million sq km of forest area, according to a 2016 report by World Bank.
Using TensorFlow, Day classified satellite images of earth’s surface into three categories— location, plant type and deforestation. Once the image classification model was created it was optimised using Open Vino. It was then ran on Intel’s Movidius Neural Compute Stick (a USB device for building and training AI models on single board and regular PCs and doesn’t require internet access). NCS allowed Rosemarie Day to monitor earth's surface in real time in areas with power constraints. Her study was carried out in Amazon rainforests.
Lack of clean drinking water is another global scourge and accounts for millions of deaths every year, according to World Health Organisation. Intel software developer Peter Ma developed a solution called Clean Water AI which can identify bacteria in water using pattern recognition and machine learning.
Ma took a digital microscope and connected it to a modestly powerful Ubuntu based laptop with Intel’s Neural Compute Stick connected to it. The entire system cost less than $500. The neural network at the heart of the system was able to successfully determine the shape, colour, density, and edges of the Escherichia coli (E. coli) and the bacteria that causes cholera.