International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 12 Issue: 06 | Jun 2025
p-ISSN: 2395-0072
www.irjet.net
Food recognition and calarie measurement using machine learning Madhuri Maruti Mangutkar1, Prof. A. S. Garande, 1 PG Student, SPPU
2 Prof. ZEAL COLLEGE OF ENGINEERING & RESEARCH NARHE, PUNE - 411041
---------------------------------------------------------------------***--------------------------------------------------------------------enables models to stay pertinent and efficient in the midst of Abstract – A cutting-edge technical advancement that evolving circumstances.
blends artificial intelligence and computer vision is the Food Recognition System (FRS). Food Recognition System is an effective tool for automating the identification and classification of different food products using image analysis in a world with a great deal of culinary diversity. We suggest using the very accurate deep learning model VGG30 for food recognition. Convolutional neural networks (CNNs) like VGG30 are made especially for classifying images. It can accurately identify a variety of food items because it was trained on a dataset of food photos. In order to use VGG30 for food recognition, a collection of food photos must first be produced, with each image labelled with the type of food it depicts. The preprocessed dataset can then be used to train the VGG30 model. Because of its deep architecture, it can capture the fine details of many food products and perform well even in difficult situations with changing backgrounds, lighting, and angles. Second, because of its scalability, VGG30 can manage big datasets with ease, which makes it appropriate for applications where there may be a significant volume and variety of food photos. Furthermore, the model gains from transfer learning, which minimizes the requirement for labelled food data and expedites development. Its interpretability makes it easier to comprehend how it identifies food items, which improves openness and confidence. Furthermore, VGG30 is a dependable option for precise food recognition across a variety of domains because to its broad community support and cutting-edge performance. Lastly, because of its versatility, it can be adjusted to particular jobs or preferences, which makes it a perfect fit for the Food Recognition System.
B. Long-Tailed Distribution A long-tailed distribution is a statistical concept that shows a notable accumulation of events in the tail section of the distribution. This is where unusual events or extreme values are more common compared to a standard distribution. Unlike a normal distribution, where data points are concentrated around the mean and taper off slowly, a long-tailed distribution displays an extended and frequently dense tail, signifying a greater occurrence of rare events. This distribution pattern is widespread in different practical situations, such as income distribution, web traffic, and the prevalence of uncommon diseases. C. Food Recognition Within the field of computer vision and artificial intelligence, food recognition is a fascinating and quickly developing field. It entails creating models and algorithms that can recognize and classify different foods that are portrayed in pictures or movies. Because social media and smartphones are so widely used, there is an increasing need for food-related material to be shared, which makes automatic food recognition both a technological difficulty and a useful requirement. This cutting-edge field aims to mimic human visual perception and allow computers to discriminate between different dishes, ingredients, and culinary traditions through the use of state-of-the-art machine learning techniques like deep neural networks.
II. LITERATURE REVIEW Research by Jianping He [1] et al. shown the exceptional efficacy of deep learning methods in a range of picture-based nutrition assessment applications. Food portion proportions and food categorization are two examples of these uses. But current approaches focus on single activities, which causes problems when numerous tasks need to be completed in parallel in real-world circumstances. The authors used a multi-task learning strategy to address this issue, training both the classification and regression tasks at the same time through the use of soft parameter sharing based on L2-norm. To improve food portion size estimation accuracy, the authors also suggested combining cross-domain feature adaption with normalization. The authors' results outperform the current approaches for portion assessment in terms of mean absolute error and classification accuracy, showing great promise for the advancement of picture-based nutritional evaluation.
Key Words: Continual Learning, Long-Tailed Distribution, Food Recognition
1.INTRODUCTION A. Continual Learing In the domains of artificial intelligence and machine learning, the concept of continuous learning, also referred to as lifelong learning or incremental learning, presents a paradigm shift. Its main goal is to tackle the challenge of adapting models to changing data distributions over time. Unlike conventional machine learning methods that assume a fixed dataset, continual learning aims to empower models to continuously learn from new information while retaining knowledge gained from past experiences. This dynamic framework is especially vital in practical applications where data is non-stationary, as it
© 2025, IRJET
|
Impact Factor value: 8.315
|
ISO 9001:2008 Certified Journal
|
Page 106