Kuler achar

Published: 10 September 2024| Version 1 | DOI: 10.17632/7zwbz8gdr9.1
Contributors:
Biswajit Balo, Tanmay Sarkar

Description

*Project Title:* Good and Bad Classification of Kuler Achar Using OnePlus Nord CE 2 Lite 5G Mobile Camera *Description:* The project "Good and Bad Classification of Kuler Achar" is focused on developing a system for automatically classifying kuler achar (Indian jujube pickle) into "good" and "bad" categories based on visual characteristics. The dataset includes over 500 images of kuler achar, equally divided between good and bad samples. These images were captured using the OnePlus Nord CE 2 Lite 5G mobile camera, ensuring high-quality visual data for training a machine learning model. The images were taken against a white background in daylight conditions to ensure consistency, clarity, and minimal environmental interference. *Dataset Composition:* - *Good Samples (High-Quality Kuler Achar):* The dataset contains more than 250 images of high-quality kuler achar. These samples exhibit desirable characteristics, such as consistent texture, proper fermentation, vibrant color, and no visible signs of contamination or spoilage. These images represent the positive class for the classification model, showcasing what is considered high-quality and safe kuler achar. - *Bad Samples (Low-Quality Kuler Achar):* More than 250 images depict low-quality or spoiled kuler achar. These samples may exhibit visual signs of spoilage, such as mold growth, discoloration, uneven texture, or over-fermentation. These images form the negative class for model training, helping to identify unwanted and unsafe products. *Data Collection Setup:* All images were captured using the OnePlus Nord CE 2 Lite 5G mobile camera, which provides high-resolution imaging capabilities to capture fine details. A white background was chosen to create a neutral, non-distracting environment that highlights the key features of the kuler achar. Daylight conditions were used for consistent and natural lighting, allowing the camera to capture the true color, texture, and visual quality of the samples. *Image Characteristics:* The dataset features a wide range of kuler achar samples, varying in appearance due to different stages of fermentation, levels of spoilage, and types of ingredients. This diversity ensures that the classification model can generalize across different variations in the appearance of kuler achar, allowing it to effectively distinguish between good and bad samples. *Data Annotation:* Each image in the dataset is carefully labeled as either "good" or "bad" based on its quality. These annotations serve as the ground truth for the machine learning model, providing the necessary data for training, validating, and testing the classifier. *Data Preprocessing:* To ensure the model receives optimal input data, several preprocessing steps are applied: - *Resizing:* All images are resized to a uniform resolution to maintain consistency across the dataset.

Files

Categories

Biological Classification, Characterization of Food

Licence