Fish is a very nutritious dish that is consumed worldwide as a complete meal. This causes a rise in fish production and storage. Freshness of fish that is stored in ice boxes and deep freezers can degrade very quickly. A stale fish can cause great harm to a human ingesting it as it may carry many diseases. This paper aims to present a hybrid model that can recognise Puntius (commonly Puti Fish) as fresh or stale by using an image and certain values graded on scale of 10 like color, texture, etc. This mainly aims to a non-destructive approach to classify fishes as fresh or stale. The model comprises of a CNN for processing images and a Dense network that extracts information from the numerical data. These features are combined and are then further passed onto another dense neural network that performs the final classification task. We were able to achieve 96–98% accuracy with this model.
Keywords
Food safety
Image analysis
Artificial intelligence
Food technology
AI University Montana
AI For Research Student Community
Comments