Medicinal capsule image dataset
– Overview
This dataset was created for the purpose of evaluating the performance of visual inspection algorithms based on image processing. It consists of normal and anomaly images of medicinal capsules. You can train inspection algorithms to classify normal and anomaly capsules, as well as perform quantitative evaluations of inspection performance on capsule products by using this dataset.
This dataset can be used for both supervised and unsupervised learning approaches. For supervised methods, we assume k-fold cross-validation, where the dataset is divided into k subsets: one subset is used for evaluation and the remaining k–1 subsets for training. For unsupervised methods, you are encouraged to split the normal images into training and testing subsets before use.
When you use this dataset in publications such as conference papers or journal articles, please cite the following URL or reference.
URL:http://isl.sist.chukyo-u.ac.jp/archives/capsule
＜Reference＞ 村上尚生，平松直人，小林大起，秋月秀一，橋本学，固有空間における情報合成に基づく高リアリティ不良品画像生成，精密工学会誌，Vol.90，No.8，pp.662-668，2024.
– Released Data
This dataset consists of images captured by an industrial camera with a resolution of 1920 x 1080 pixels and 3-color channels. Each image was center-cropped to 512 x 512, converted to grayscale, and resized to 128 x 128. All images are provided in PNG format.
Download: Medicinal capsule image dataset (10.3 MB)
■Normal（600 images，5.0MB） ■Anomaly（600 images，5.0MB）
– Directory Structure
dataset
|--Normal
| |--001.png
| |--002.png
|
|--Anomaly
|--001.png
|--002.png
This dataset consists of two classes: Normal and Anomaly. Each class contains a set of grayscale images. For supervised learning methods, we assume the use of k-fold cross-validation, where the dataset is divided into k subsets — one subset is used for evaluation, and the remaining k–1 subsets are used for training. For unsupervised learning methods, please split the normal images into training and testing subsets before you use.