The internet is a transformative force in information technology, with over 66.2% of the global population connected. Social media platforms like Instagram have revolutionized communication and data sharing, becoming crucial tools for various industries, including food. Food photography on Instagram, commonly known as “foodstagramming,” reflects evolving eating habits where food represents personal identity and lifestyle. This study explores Instagram user-generated content (UGC) to identify food trends in Bandung, a city renowned for its culinary diversity. By employing Convolutional Neural Networks (CNNs), this research classifies food images and derives insights that can inform menu diversification, marketing strategies, and business decision-making. The analysis focuses on three predominant food categories: fried food, noodles, and soup, gathered from Instagram hashtags #kulinerbandung. The CNNs model is trained using 9,339 image data. this model produces an accuracy value of 0.9994 and a validation value of 0.8412. This shows that the CNNs model with local food image data especially Bandung culinary can be recognized very well. This research results in the importance of utilizing UGC to analyze consumer preferences to improve business competitiveness. The findings also serve as further research on local culinary trends and propose the implementation of deep learning to analyze consumer preferences.