News classification
Contact us
- Add: No. 9, North Fourth Ring Road, Haidian District, Beijing. It mainly includes face recognition, living detection, ID card recognition, bank card recognition, business card recognition, license plate recognition, OCR recognition, and intelligent recognition technology.
- Tel: 13146317170 廖经理
- Fax:
- Email: 398017534@qq.com
By See Food Inc developed hot dogs recognize app "is Not a hot dog" (Not Hotdog) triggered a heated discussion, AI community, however a seemingly simple XiXueXing app involves the AI technology is very complicated, although only a small at the application level, AI milestone in the history of mankind, the same product, 36 kr before related reports.
The popularity of Not Hotdog has sparked another wave of food recognition in the AI community, which has recently been studied in the MIT Computer Science and Artificial Intelligence Laboratory. The MIT team intends to use the food recognition system to identify ingredients for recipes from food production video. MIT team pic2recipe (image to the receiver) system using computer neural network to determine the kinds of food in food pictures on social networking, which can be further analyzed the uploader health habits and dietary preferences.
Pic2recipe system developed by using the Swiss scientist in 2014 Food Food - 101 Data Set identification algorithm, using the database of 101000 Food pictures, and these pictures with CSAIL Recipe1M cross reference part of the database Data. The data for Recipe1M database is mostly pulled down from popular recipe sites such as All Recipes and Food.com.
Currently, the technology is still a long way from full maturity, and the accuracy of the current system recognition is only about 65%. The biggest bottleneck currently encountered in the project is the image itself. Joint research and development person Nick Hynes said, people, food photos were taken food are influenced by the camera mode, including Angle, distance, display and lighting and other factors can contribute to the different identification results. When the same food appears in different recipes, the recognition error rate of the system increases.
The system is better at recognizing baked goods.
Food has a trillion-level market, with a wide range of vertical sectors, ranging from retail to restaurants to social industries. Under the food content website, the kitchen, gastronomy and other enterprises have been invested, and believe that the technology developed by MIT is very large if the technology can be applied to the scene.