Machine Learning-Based Automatic Classification of Gastrointestinal Categories from Food Images

The meal type, calorie count, and weight are all estimated using a smartphone app. The created software classifies the user as underweight, normal weight, overweight, or obese based on many medical criteria like BMI, etc. The software also sorts foods according to their GI. The software estimates the user’s weight and caloric intake using cloud-based ML. Eating recommendations for optimal nutritional management is another area the app will expand its utility for. The recommendations will take into account a number of underreported dietary characteristics, such as body mass index (BMI), age, sex, food type, eating habits, etc. This study will not only help medical experts make correct diagnoses, but the technology will also be able to supply individuals with self-eating instructions. In areas where healthcare equipment is lacking and expertise is few, this breakthrough will provide a significant answer for the pharmaceutical and wellness sectors. Dietary management programs will also find this product useful for demonstrations and lessons. Reliable scientific tracking of intake categorisation systems by food chews event identification in the portable context is a medically important aspect of this work. The suggested AdaBoost with Random Forest approach achieves much higher accuracy than the other methods when compared to their respective performances. J48-DT delivers good results, KNN improves accuracy, and Naive Bayes performs adequately. The suggested strategy, on the other hand, has an astounding $94.1 \%$ accuracy.