The nomination to Best Paper Award at the upcoming CIARP conference belongs to a team with IEETA signature. The team is composed by João Rodrigues (IST), Osvaldo Pacheco (IEETA), and Paulo Correia (IST). CIARP will be held in Coimbra this November 2023. The conference is focused on disruptive innovation on pattern recognition, computer vision, artificial intelligence, and related domains, is set to host a gathering of innovative minds.
The nominated paper, titled “Seabream Freshness Classification using Vision Transformers” introduces a groundbreaking automated system for classifying fish freshness. This non-invasive approach utilizes fish images to categorize freshness into four distinct levels: very-fresh, fresh, not-fresh, and spoiled. Notably, this research also unveils a novel seabream image dataset, soon to be publicly available for academic and scientific purposes.
What sets this work apart is its comprehensive dataset, including metadata such as manually generated segmentation masks for fish eye and body regions, along with the time since capture. The process involves initial image segmentation to isolate the fish’s eye region, followed by freshness classification based on unique eye characteristics. A remarkable aspect of this system is the incorporation of transformers, specifically Segformer for segmentation and Vision Transformer (ViT) for feature extraction and freshness classification. This marks a pioneering use of transformers in the assessment of fish freshness.
This nomination is a testimony to the team’s commitment to innovation and pushing the boundaries of technology in practical applications.