Research And Application Of Diet Recommendation Algorithm For Fitness Enthusiasts Based On Deep Learnin

In recent years,with the rapid development of China’s economy,more and more people are participating in fitness exercises to prevent diseases,strengthen their bodies and change their figures.However,due to objective constraints,most ordinary fitness enthusiasts are unable to customize a scientific diet plan for themselves.Therefore,it has become a relevant topic to recommend personalized meal plans for users based on their nutrient intake needs and dietary preferences.In this paper,a personalized diet recommendation model is studied based on deep learning technology,and a diet recommendation algorithm for fitness enthusiasts is designed to recommend a nutritional meal plan that satisfies both dietary preferences and nutrient requirements with food diversity,taking into account the dietary nutritional needs of fitness enthusiasts.The main work of this paper is as follows.(1)A Multi-Modal Personalized Dietary Recommendation Model Fusing User’s Visual Preferences(MMFV)is proposed.First,to address the important influence of visual features on the dietary recommendation task,images of food are added to the model to complement the features,and a user visual preference extraction module is designed to model the user’s visual preferences based on the Query-Key-Value attention mechanism using the user’s history of clicking food images.Then,on multi-modal feature fusion and interaction,a new feature fusion and interaction module is designed to address the shortcoming that the traditional stitching+multi-layer perceptron approach cannot distinguish feature interaction importance,to mine feature interaction importance using the proposed multi-head bit-wise attention mechanism,and to update the higher-order fusion representation of features according to their interaction importance.In this paper,experiments are conducted on Fit Foods dataset and Movielens-1M dataset,and the experimental results show that the MMFV model has optimal performance and good efficiency compared with other benchmark recommendation models.(2)A personalized fitness diet recommendation algorithm was designed based on bodybuilders nutritional catering theory and MMFV model.Firstly,the calorie requirement of the user for the meal is calculated based on the user’s gender,age,weight,physical activity level and weight target,and the calorie ratio provided by each nutrient is used to calculate the intake requirement of each nutrient.Then the meal plan generation method is designed according to the balanced diet model and MMFV model,so that the generated meal plan can combine dietary diversity and user’s dietary preference.In order to determine the amount of food consumption in the meal plan,this paper plans and solves the multi-objective optimization problem based on Multiple Objective Particle Swarm Optimization(MOPSO),and adds a calorie check mechanism to address the problem that the calorie intake of some solutions in the Pareto solution set does not meet the user’s fitness goal.The calorie checking mechanism is added to restrict the non-dominated solutions and guide the population development.In this regard,simulation experiments are conducted in this paper,and the experimental results show that the fitness attainment rate of the Pareto optimal solution set is significantly improved after the algorithm improvement,and the error of each nutrient is less than 1% in multiple experiments.(3)Based on the proposed diet recommendation algorithm for bodybuilders,a personalized diet recommendation system for bodybuilders is designed and implemented.The system is based on B/S architecture.The front end uses Vue + element UI and the back end uses flask framework to realize the function of personalized nutrition recipe recommendation.Firstly,the reference amount of nutrients in the current meal of the day is calculated according to the user’s information,diet and exercise,and the MMFV model is used to generate a diet recommendation scheme that meets the user’s preferences for the user.Finally,the optimal amount of each food in the scheme is solved by using the designed multi-objective optimization algorithm of nutritional catering and recommended to the user.