Consumers are becoming more demanding when it comes to food. They are increasingly sensitive to reports of safety incidents and expect high food quality and transparency along food chains.
The background to this is the recurrence of scandals, as recently in meat processing, but also the need for sustainable food. In this sense, sustainable food quality results from two aspects that complement each other in the meat industry:

• Species-appropriate and ethically justifiable rearing, keeping and slaughtering of animals, which contributes to a sense of quality on the part of consumers and to their own perception of a sustainable lifestyle that is compatible with animal welfare,
• Reliable food safety that not only reacts to problems and increases resource consumption through preventive measures, but is implemented sustainably and above all as proactive, predictive food safety, which also includes the goal of resource conservation.

In this regard, AI-based predictive assessment on food safety incidents also contributes to the reduction of food waste. The data collected in existing systems for tracing and ensuring food safety allow statements to be made about the origin of raw materials and enable rapid and reliable clarification. However, the approach is reactive. Problems can only be tracked as they occur. Predictive approaches that identify problems before they occur or become relevant are not supported.

The goal of the KINLI project is to conceptualize an AI-enabled predictive approach to ensure sustainable food quality, implement it in a platform, and demonstrate and validate the concept and platform in practical use cases.
This supports the sustainable design of food chains. The project focuses on the meat industry, but the concept is designed to be transferable. It will be demonstrated by linking production data and AI-based early detection of problem cases relevant to food safety, as well as detection and assessment of rearing and housing conditions using deep learning techniques.

The overall project is divided into work packages.
The first work package comprises the creation of a concept that uses AI to ensure sustainable food quality in supply chains. To this end, the user-centered development of the KINLI system will be planned and controlled, taking into account all relevant stakeholders as the foundation course for the technical implementation.
Subsequently, the necessary technical platform to support the developed AI services will be created.
For this purpose, first the data model, then the data platform and finally the data management will be implemented. In the third work package, the developed concept will be prototypically implemented in AI services. The AI solution is validated with the help of two concrete use cases within the meat processing industry. This is followed by a detailed conceptual design as well as data collection and preparation for the implementation of machine learning models. In the fifth work package the evaluation of the system takes place. The system is examined with regard to the fulfillment of technical requirements, user-friendliness and practical suitability.