NutriFit-AI: Nutrition System of Vision Based Calorie Estimation and Fitness Welfare | IJCT Volume 13 – Issue 3 | IJCT-V13I3P13
International Journal of Computer Techniques
ISSN 2394-2231
Author
Prof. Tanvir H. Sardar, Varikuti Dayananda Reddy, Dudekula Noor Mohammad, Chillara Sree Charan Karthik, Kalyam Chandu Reddy
Abstract
Due to the rising rates of obesity, physical inactivity, and improper dieting, the demand to have available and intelligent wellness solutions has increased in direct proportion. Although there are numerous fitness and nutrition apps nowadays, most of them are single-use or need manual data entry or professional supervision and adherence and long-term are minimized. In addition, people are often lacking instant feedback of exercise performance and tracking of calories intake usually seems tedious and inaccurate. In this paper, we introduce NutriFit AI which is a multimodal intelligent fitness and nutrition system that combines computer vision-based exercise analysis with vision-based calorie estimation on a single web platform. The vision models associated with the proposed system are deep-learning-based, which means that food images are processed to give approximate values of calories and macronutrients, rendering the process of dietary tracking easy and allowing the removal of manual food recording. Simultaneously, human pose estimation is implemented to the workout videos to assess the motions of body joints and make it possible to count the repetitions of an exercise automatically, without relying upon wearable sensors. The system is provided through the use of a Flask-based server, OpenCV, and MediaPipe as pose estimators, a lightweight SQLite database to store history and progress trends of users. It has been experimentally demonstrated that the system is highly reliable in repetition counting of popular bodyweight activities, and provides useful nutritional estimations of regular meals. Although the system is not claimed to be clinical in the precision, it is useful in helping to perform self-monitoring on a daily basis and long-term fitness awareness. NutriFit-AI will prove the viability of a low-cost, user-friendly, and well-integrated AI-based virtual fitness assistant that could be used in the context of real-life personal wellness management.
Keywords
Artificial intelligence (AI), Computer vision, Pose detection, Nutrition detection, Fitness tracker, Vision-based calorie counting, MediaPipe..
Conclusion
The paper presented NutriFit-AI, a multimodal system enhancing the access to personal fitness and nutrition monitoring with the artificial intelligence and computer vision applied. The motivation behind the work is the challenge people experience when it comes to exercising on a regular basis and keeping track of their diets without the help of a professional. Most of the available tools use manual data or do not consider dietary and activity individually, which makes them less useful in the long term.
The suggested site integrates the evaluation of the BMI, the use of food photos to determine the nutritional value, and the analysis of poses during exercise into one web application. An AI interface is arranged through a Flask server that organizes the AI services of the calorie knowledge and body recognitions, and a database which records the previous data. This integration allows its users to see the history of trends and get a more meaningful picture of how the habits change in time. Intended use was found to be reliable in an experimental application. Computation of BMI was held constant, nutritional products provided feasible estimations and counting of repetition performed successfully when the subject was in full view. The integrated dashboard was particularly useful in displaying diet and activity data side-by-side, which motivates a person to reflect on themselves.
All in all, the case of NutriFit-AI demonstrates that valuable digital wellness care is achievable by using low-cost infrastructure and widely accessible cameras. The system offers a realistic basis of achieving long-term intentions of engaging in healthy lifestyle behaviours, which is achieved by reducing the manual effort and focusing on the interpretable feedback.
References
[1] M. Kaushik et al., Posture Correction, Real-Time Exercise Tracking and Feedback M. Kaushik et al. in Proc. 4th Int. Conf. Comput. Graph. Reality (CGR), Virtual, 2024, pp. 1–6.
[2] X. Fan et al., “Combining local appearance and holistic view: Dual-source deep neural networks in human pose estimation, in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), Boston, MA, USA, 2015, pp. 1347–1355.
[3] J. Zhu et al., “A Pose Estimation Approach in Dynamic Scene using YOLOv5, Mask R-CNN and ORB-SLAM2, in Proc. 7th IEEE Int. Conf. Signal Process. (ICSIP), Suzhou, China, 2022, pp. 456–461.
[4] G. Taware et al., “AI-based Workout Assistant and Fitness guide, in Proc. 6th Int. Conf. Comput. Commun. Ubiquitous Eng. Appl. (ICCUBEA), Pune, India, 2022, pp. 15.
[5] G. Dsouza et al, Smart gym trainer based on Human pose estimation, in Proc. IEEE Int. Conf. Innov. Technol. (INOCON), Bangalore, India, 2020, pp. 1–4.
[6] L. Jiang et al., “DeepFood: Food Image Analysis and Dietary Assessment with Deep Model, IEEE Access, vol. 8, pp. 4747747489, Mar. 2020.
[7] S. Naritomi et al., CalorieCaptorGlass: Food Calorie Estimation by Real Size using a See-Through Glass and a Kinect in Proc. IEEE Virtual Reality Workshops (VRW), Atlanta, GA, USA, 2020, p. 741- 742.
[8] P. Kumar, S. Senapati, etc. Calorie Estimation of Food and Beverages with Deep Learning, in Proc. 7th IEEE Int. Conf. Comput. Methodol. Commun. (ICCMC), Erode, India, 2023, pp. 324–329.
[9] S. Ibrahim et al., Deep Learning Framework to Detect and Track Calories in Food in 2023, in IEEE Int. Conf. Comput. (ICOCO), Langkawi, Malaysia,2023,pp.397–401.
[10] S. Jayanthi, V., 2023, 9 th Int. Conf. Smart Comput., Using Faster R-CNN and Mask R-CNN, Food Image Analysis and Calorie Prediction. Commun. (ICSCC), Kochi,India2023,pp.83–89.
[11] P. Yarde, D. Bordoloi, R. M. Chavan, V. Vekariya, H. Patil and L. Natrayan, “A Deep Learning Neural Network-based System for Food Recognition and Calorie Estimation,” Proc. 3rd Int. Conf. Innov. Ind. Appl. (ICIMIA), Bengaluru, India, 2023, pp. 1551–1558.
[12] P. Poply and J. A. Arul Jothi, “Refined image segmentation for calorie estimation of multiple-dish food items,” in Proc. 2021 Int. Conf. Comput., Commun. and Intelligent Systems (ICCCIS), Greater Noida, India, Feb. 19–20 2021, pp. 682–687.
[13] R. Ruede, V. Heusser, L. Frank, A. Roitberg, M. Haurilet and R. Stiefelhagen, “Multi-Task Learning for Calorie Prediction on a Novel Large-Scale Recipe Dataset Enriched with Nutritional Information,” in Proc. 2020 25th Int. Conf. Pattern Recognit. (ICPR), Milan, Italy, 2021,pp.1555–1562.
How to Cite This Paper
Prof. Tanvir H. Sardar, Varikuti Dayananda Reddy, Dudekula Noor Mohammad, Chillara Sree Charan Karthik, Kalyam Chandu Reddy (2026). ^PAPER_TITLE^. International Journal of Computer Techniques, 13(2). ISSN: 2394-2231.
