Coca-Cola uses TensorFlow as a marketing tool in its customer loyalty programs. Coca-Cola printed the 14-digit code on the back of the bottle cap, and created a machine learning system that can easily identify the numbers and successfully create the proof of purchase by using TensorFlow. Up to now, the promotion of 16 has been promoted, and the code has been scanned more than 80,000 times.
2. In agriculture
The Dutch company Connecterra used TensorFlow to create a wearable device called "Cow Bracelet". Through machine learning, Connecterra can diagnose problems in the early stage and provide suggestions to help farmers maintain the health of their cows.
This set of sensors, which can be connected to the mobile application "Ida", can analyze the movement of cows and provide farmers with key information to know when cows get sick, when they need special care and what kind of care they need, so that cows can produce better in a healthy state.
3. In the field of telecommunication services.
China Telecom Business Hall APP builds a neural network to learn how to identify telecom prepaid cards. Finally, it can be easily identified in the case of backlight, jitter and serious digital occlusion. The success rate of model training is 99.3%, and it takes 0.05 seconds to complete a recognition.
Today, in rural areas/elderly people/second-tier cities, 35% of people are still recharging their prepaid cards, and on average, more than/kloc-0.0 million people recharge their prepaid cards through mobile apps every day. By taking photos to identify prepaid cards, the neglected demand in the Internet era is solved, and AI really benefits the people.
4. In the application of mobile terminal.
Kika made an embedded smartphone input method through TensorFlow Lite, which reduced the memory occupation by nearly 50%. The response speed is doubled, and the accuracy of speech comprehension is also upgraded.
Sina Mobile is another example. They used TensorFlow to create a system that allowed their users to identify many car brands by taking photos. Using TensorFlow, the volume of their model is reduced to a quarter, the calculation efficiency is improved by four times, and the recognition accuracy is improved by 85%.
5. Other examples
Build your own AI and machine learning platform with TensorFlow. For example, JD.COM, Tencent, Ali, Netease, Meituan, Xiaomi and other companies. They use TensorFlow to solve various problems, from computer vision, natural language processing to recommendation prediction and so on.