Application of AI-Helped Image Classification of Fish Images: An iDigBio dataset example

• DOI: 10.3897/biss.7.112438
• Publication Date: 2023-09-11T06:30:20Z
• Authors: Bahadir Altintas; Yasin Bakış; Xiojun Wang; Henry Bart
• Keywords: Contextual image classification; Table (database)

Artificial Intelligence (AI) becomes more prevalent in data science as well areas of computational science. Commonly used classification methods AI can also be for unorganized databases, if a proper model is trained. Most the work done on image purposes such object detection and face recognition. If an detected from image, may to organize data. In this work, we try identify images Integrated Digitized Biocollections (iDigBio) dataset classify these generate metadata use AI-ready future. The main problem museum datasets lack information images, wrong categorization, or poor quality. By using AI, it maybe possible overcome problems. Automatic tools help find, eliminate fix For our example, trained 10 classes (e.g., complete fish, photograph, notes/labels, X-ray, CT (computerized tomotography) scan, partial fossil, skeleton) by manually tagged iDigBio dataset. After training each class, reclassified models. Some results are given Table 1. As seen table, even classified identified different classes, some very similar other visually scans X-rays fossils skeletons. Those kind similarities confusing human eye results.