Enhanced Food Image Classification using CNN with Background Removal and Contrast Stretching

This research explores the utilization of CNNs to improve the classification accuracy of improvised food images. It initiates a detailed process, starting from image acquisition and moving through crucial preliminary processing steps including shrinking, removing backgrounds, and enhancing contrast using techniques like imadjust. The CNN architecture is then used to precisely detect and extract features from these previously processed images. A vast range of food items are included in the classification, from Apple Pie to Fish and Chips, with the primary evaluation criterion being the model’s accuracy. Notably, the model’s performance demonstrates significant enhancement, highlighting the efficacy of the CNN-based methodology in accurately classifying diverse improvised food images. This research contributes novel Exploring viewpoints within the domains particularly in the dynamic terrain of varied food products, of computer vision and image categorization.