Ocular Image Dataset for Eye Disease Classification and Screening Research

Dataset Overview This dataset is designed for eye disease classification and research purposes. It contains labeled ocular images representing five clinically relevant categories: Cataract, Conjunctivitis, Eyelid Disorders, Uveitis, and Normal (healthy eyes). The dataset aims to support research in computer aided diagnosis, medical image analysis, and artificial intelligence based ophthalmic applications. It is suitable for academic research, algorithm development, benchmarking, and educational use, as well as exploratory studies relevant to the medical and healthcare domain. The dataset includes a total of 2,338 samples, distributed as follows: Cataract: 544 images Conjunctivitis: 368 images Eyelid Disorders: 535 images Normal: 659 images Uveitis: 233 images All samples are organized by class label to facilitate supervised learning, statistical analysis, and comparative evaluation. Data Sources and Collection Process The dataset was compiled from multiple trustworthy and publicly accessible sources. These include curated image results obtained through Google searches, educational medical websites, and reference materials commonly used for clinical and academic learning. Informal consultations with qualified medical doctors were conducted to verify the visual consistency and clinical relevance of the disease categories during dataset curation. The dataset was assembled to capture representative visual patterns of common eye conditions rather than patient specific clinical records. No proprietary medical databases or restricted clinical systems were accessed during data collection. Ethical Considerations and Data Privacy This dataset does not contain personally identifiable information, patient names, medical record numbers, demographic attributes, or metadata that could be used to identify individuals. The images are limited to ocular regions and are intended strictly for research and educational purposes. Formal institutional ethical approval or consent documentation is not available for this dataset. This limitation is disclosed transparently to support responsible use. The dataset is intended for non commercial research, method development, and academic evaluation and should not be used for direct clinical decision making or diagnostic deployment without further validation and appropriate regulatory approval. Users are expected to follow ethical research practices, respect data privacy, and comply with relevant institutional, national, and international guidelines when using or redistributing the dataset. Intended Applications and Use Cases This dataset supports a range of applications, including: Development and evaluation of machine learning and deep learning models for eye disease classification Academic research in ophthalmology and medical image analysis Benchmarking and comparative analysis of classification methods Educational use in medical imaging, data science, and healthcare related courses Preliminary and feasibility studies for computer aided screening systems The inclusion of a Normal class enables comparative analysis and supports automated screening and early detection research. Responsible Use Statement The dataset is provided to encourage research and innovation while promoting responsible and ethical use. Any publications, models, or derivative works based on this dataset should acknowledge its sources, limitations, and intended scope. The dataset creators do not claim clinical completeness or diagnostic authority, and the dataset should be treated strictly as a research resource rather than a clinical tool. Dataset Structure and Contents After extracting the RAR file named “Eye Diseases Classification”, a main directory titled“Image Dataset on Eye Diseases Classification (Uveitis, Conjunctivitis, Cataract, Eyelid, Normal) with Symptoms and SMOTE Validation” is created. Within this directory, the dataset is organized into five class specific subfolders: Cataract Conjunctivitis Eyelid Disorders Uveitis Normal Each subfolder contains labeled ocular images corresponding to its category. The directory structure is designed to support reproducible experiments, supervised learning pipelines, and image based classification research.