Samosa

Published: 20 August 2024| Version 1 | DOI: 10.17632/z8r3pvv863.1
Contributors:
kartick mondal, Tanmay Sarkar

Description

**Data Description: Samosa Classification Dataset** **Overview:** The dataset comprises images of samosas categorized into two classes: 'Good' and 'Bad.' A total of over 500 images have been collected, capturing the variability in appearance within each class. These images were taken using a Samsung J7 mobile camera under consistent daylight conditions against a white background, ensuring uniform lighting and minimal background interference. **Data Composition:** - **Total Images:** Over 500 - **Good Samosa:** Approximately 250 images - **Bad Samosa:** Approximately 250 images **Image Details:** - **Camera Model:** Samsung J7 - **Background:** White - **Lighting:** Daylight, providing natural and consistent illumination - **Resolution:** Varies depending on the Samsung J7’s camera settings but typically around 1280x720 pixels **Labeling:** Each image is meticulously labeled as either 'Good' or 'Bad' based on predefined criteria. 'Good' samosas meet quality standards in terms of shape, color, and texture, while 'Bad' samosas exhibit imperfections such as poor shape, uneven coloring, or visible defects. **Class Definitions:** - **Good Samosa:** Characterized by an appealing appearance, consistent color, and proper shape. The samosas in this class exhibit desirable quality attributes expected in a standard product. - **Bad Samosa:** Exhibits defects such as irregular shape, inconsistent color, or visible damage. These samosas do not meet the quality standards set for 'Good' samosas. **Usage Context:** This dataset is intended for developing and testing machine learning models for image classification. It is designed to help improve automated quality control processes in food production or similar applications where visual inspection is critical. The consistency in background and lighting aims to reduce variability in image quality, focusing the classification task on the inherent characteristics of the samosas. **Preprocessing:** Images are captured under controlled conditions to minimize external variables. However, preprocessing such as resizing, normalization, or augmentation may be applied depending on the requirements of specific machine learning models. **Challenges:** - Variability in samosa appearance due to factors like size, shape, and filling consistency - Ensuring accurate labeling and quality control for 'Good' versus 'Bad' classification **Applications:** - Development of image recognition systems for quality assessment - Training of machine learning models for automated defect detection - Research in computer vision techniques applied to food quality evaluation This dataset provides a foundational resource for advancing image classification and quality assessment technologies in the food industry.

Files

Categories

Biological Classification, Characterization of Food

Licence