Enhancing the Performance of Dog Breed Classification and Food Recommendation System using Visual Geometry Group 16 in comparison with Long Short-Term Memory

The ultimate aims to improve the accuracy in dog breed classification and food recommendation systems by using Novel Visual Geometry Group 16 (NVGG16) instead of Long Short-Term Memory (LSTM). The researchers used NVGG16 and LSTM to analyze 20,981 images of 120 dog breeds from the Dog Breed Identification dataset on Kaggle. ClinCalc is used to calculate a sample size of 95% confidence and an alpha value of 0.05. The performance of the dog breed classification and food recommendation system was evaluated based on its accuracy. In comparison to Long Short-Term Memory, which scored 67.72%, the mean accuracy in classifying Dog Breeds and recommending food using Novel Visual Geometry Group 16 is significantly higher at 92.12%, according to the Results section. With an accuracy p-value of 0.036 (where p < 0.05), a statistical analysis, more precisely A substantial difference between the two approaches was shown using an Independent Sample T-Test. Conversely, with regard to loss, a similar statistical difference was seen. As a result, this study highlights the superiority of the Novel Visual Geometry Group 16 method over the Long Short-Term Memory algorithm in terms of accuracy for both Dog Breed Classification and Food Recommendation System.