If you believe all cows look the same more or less, you might want to read on. Cows in some districts of Punjab are currently part of an intriguing project that involves machine learning telling one cow apart from the other. Every cow and buffalo are distinguishable as agritech startup Mooofarm, the brains behind the algorithm, claims to have 95% accuracy in distinguishing one from the other. The algorithm only requires pictures of each cow or buffalo from different angles, backgrounds and lighting to identify unique features of the animal.
Founded in 2017 by Aashna and Param Singh, Mooofarm is an agritech firm dedicated to finding technical solutions for dairy farmers by helping them keep their cattle healthy and increase their income by improving the milk quality.
The startup has built a digital ecosystem of cattle through its mobile app that leverages data analytics to provide farm and cattle management solutions to dairy farmers such as digitizing the life cycle of their cattle, connecting them to input suppliers for fodder and to veterinarians for emergencies, among other services.
It’s no bull: Technology prevents frauds, diseases
So, what could be the real-world applications of such technology? “This technology can help solve the fraudulent claim settlement issue in the cattle insurance sector wherein the farmer claims (sends ear/ear tag) of an insured cattle, whereas the death has been of uninsured cattle,” Aashna Singh, co-founder, Mooofarm, told ET.com.
The numbers representing livestock insurance in India are dismal and plummeting. According to a reply given in Lok Sabha on March 13, 2018, 14.80 lakh animals were insured in 2014-2015. Since then, the number has dropped to 7.44 lakh in 2016-2017.
Further, current practices of claiming insurance for cattle are morbid. To claim insurance, for most animals, a death certificate from a veterinarian, RFID (Radio frequency identification chip, inserted in the front left leg of the animal) and a post-mortem report are required. However, for cows, it is imperative to submit the chopped off part of the deceased animal’s ear, which had been earlier punched with an ear tag carrying a unique number when the animal is bought.
“Mooofarm’s facial recognition solution has the capability to replace this method altogether,” said Aashna.
The machine learning algorithm is fed with 20-30 pictures of each cow taken from different angles, different backgrounds, and different lighting. 1000s of such pictures were augmented and tested to train the model. This is similar to training the FaceID algorithm of an iPhone where you rotate your face in front of the phone’s camera to identify your face.
The pictures of two cows may look the same to us, but they have distinguishing physical features all across their face, including muzzle and eyes, which the ML catches. Therefore, this also helps in classifying the cattle in terms of ages, breeds and other categories.
“Pictures with different lightings and backgrounds are taken to train the algorithm so that later when a farmer takes a cow’s picture in low lighting, the animal’s unique features are still recorded,” the London School of Economics alumni said, adding that they are further working on improving the accuracy of the model.
This project won a $30,000 prize grant from the World Bank’s Agriculture Insuretech Innovation Challenge this year.
Interestingly, Mooofarm is using the smartphone cameras for another significant cause. Collaborating with technology giant Microsoft, the firm is helping dairy farmers detect mastitis, a potential fatal mammary gland infection affecting a cow’s udder. This leads to decreased quantity of milk and risking its quality by causing compositional changes.
“Mastitis is a disease of the udder that leads to half-a-billion dollar loss per year,” Aashna said, adding that the Mooofarm team won a grant to build a Mastitis detection technology using Microsoft technologies.
She added that the disease can be detected by visible and non-visible symptoms. The startup is currently in the process of building an ‘image labelling’ technology, in association with Microsoft. Here, the farmer will have to take images of the cow’s udder and milk it produced, and the ML algorithm will then detect whether the cattle is affected by the disease or not.
Digitizing the dairy industry
While the Cattle Facial Technology and Mastitis Detection Technology are recent additions to Mooofarm’s platform, the company’s aim since its inception has been to empower smallholder dairy farmers.
In its early stages, the startup would provide farmers door-to-door training, set up village level awareness campaigns, connecting them to veterinarians and giving them training in dairy practices on how to better take care of their cattle, the right nutrition, educate them about diseases, etc. The firm had a team of village level entrepreneurs who would again go door-to-door to make sure the taught practices are implemented properly.
“During the training, we realized that there’s this huge need for the influx of technology and innovation, because there was no record keeping or follow up on something as basic as the breeding cycle of a scoured buffalo by farmers,” Aashna said, adding that the first thing they did in the Mooofarm app was digitizing the breeding cycle of cattle, so farmers will know their cattle is in heat.
The farmer will have to upload the basic inputs of his cattle to get real-time alerts such as the right time to inseminate the cattle and the right time to get a pregnancy diagnosis. The app also has an inbuilt e-commerce platform which allows farmers to connect with input suppliers (food, fodder seeds) and other service providers such as insurance and veterinarians.
The team’s village level entrepreneurs go to each doorstep to ensure that the app is installed and implemented fruitfully. Interestingly, when the app was first deployed, many farmers came forward with other issues and helped the Mooofarm team upgrade the app. These included making an entry for the medicines fed to the cattle, the kind of insemination done, the amount of fodder given to the cattle, etc.
For each of these inputs which enhanced the application, the farmers were given Mooo points, a form of redeemable loyalty points in exchange of internet recharge, mineral mixture packets and other dairy inputs.
Gurucharan Singh, a farmer from Chamaaroo, Punjab, possessed the incorrect knowledge about the amount of feed needed for his cattle. He thought the more feed given to the cow, the more milk production, which resulted in him spending a huge amount on cattle feed.
Mooofarm’s trainers removed this misconception and taught Gurucharan and other farmers the formula to calculate the accurate amount of feed for each cattle. The farmer was able to reduce the feed amount to 33 kg per day, saving Rs 6390 per month.
Mooofarm began as Project Mooo in 2017 which was an initiative by UDAY, a skill development company, founded by Param in 2014, after he sold his Australia-based founded company. UDAY, partnering with NSDC (National Skill Development Corporation), provides extensive skill programmes to women, youth, and farmers to help them become self-sustained.
The focus soon expanded to redefine the dairy industry in the country by providing solutions to not only increase the milk production, but also increase the income of smallholder dairy farmers. After 18 months of the pilot project, Aashna and Param turned it into a full-fledged company in July this year.
“Our immediate goal is to impact 2 lakh farmers by end 2020 and our vision is to enhance 1 million lives by 2022,” Aashna said. The company is a B2B entity and ties up government agencies and corporates who want to spend their corporate social responsibility budgets for farmer welfare. “We’ll soon offer a SaaS model,” she added.
Mooofarm is currently present in Sangrur and Patiala districts of Punjab (starting Ludhiana and Moga soon), Hamirpur and Hardoi district of UP, Nagpur Amravati and Wardha district of Maharashtra, and Dausa district of Rajasthan.
Let’s block ads! (Why?)