On 25 july, press releases (www. Thepaper. Cn) learned from the chinese agricultural institute that the academy's oil quality chemistry and processing team had made a major breakthrough in the field of oilseed quality testing。
Through the integration of computer visual and artificial intelligence techniques, the team successfully constructed a high-quality image database and model library covering a wide variety of oilseeds, the first online real-time second-scale testing of the quality of oilseeds. Research results have been published in international authoritative journals such as food chemistry and trends in food science & technology。
According to the author of the paper, weifang, a researcher at the ministry of agriculture and rural affairs' oil processing focus laboratory, traditional methods of testing the quality of oilseeds rely on sophisticated instruments and laboratory analysis, not only for the destruction of samples, but also for time and effort to meet large-scale, real-time testing needs. To that end, the team proposed an innovative “photography-to-measure” programme, using computer-based visualization techniques based on in-depth learning, training in a light quantitative depth learning model and developing a seedvision software suitable for computer and mobile phones to detect oilseeds, protein, chlorophyll and total phenol content only for uploading and within 10 seconds, with an accuracy rate of over 88 per cent and an average error of less than 5 per cent, providing technical support for real-time online monitoring of the quality of oilseeds and oil crops such as peanuts and beans。
This breakthrough will significantly increase the efficiency of quality testing of oil crops, promote the development of agricultural intelligence and provide critical technological support for food and oil quality security and industrial efficiency。
