Fresh Fish Classification Using HOG Feature Extraction and SVM

Fish freshness is a significant factor in deciding if the fish is still safe for consumption or not. When the fishes are not adequately inspected before distributing them to the consumer, it can cause Scombroid and Ciguatera infection for human consumption. Additionally, fishes that are mistakenly included in distribution could lead to potential waste and dissatisfaction from consumers. Using the fish common in the local fish market in the Philippines, this study is focused on developing a way to classify the fish based on their freshness and specie. With the Histogram of Oriented Gradient and Support Vector Machine, the system could recognize the fish’s specie. The same method would also be used to recognize the fish’s freshness based on the eye of the fish. Raspberry Pi microcontroller is also used to implement this experiment. The research used 20 sample fishes, and 18 of them were correctly recognized by the system resulting in a 90% accuracy.