Good and Bad classification of cooked pasta

Good and Bad Classification of Cooked Pasta
This project, titled “Good and Bad Classification of Cooked Pasta”, aims to develop an image classification system that distinguishes between properly cooked (good) and improperly cooked (bad) pasta. The dataset consists of approximately 2000 images, equally divided into two classes: 1000 good samples and 1000 bad samples. All images were captured using a Realme C25Y smartphone camera, providing high-resolution images suitable for machine learning applications. The photographs were taken under controlled lighting conditions to maintain clarity, consistency, and accurate feature representation.
Dataset Composition
Good Samples (Properly Cooked Pasta):
The dataset includes around 1000 images of properly cooked pasta. These images display pasta with uniform texture, smooth surface, consistent color, and well-maintained shape and structure. There are no burnt edges, undercooked portions, or discoloration. These samples represent the positive class and help the model learn the characteristics of well-cooked pasta.
Bad Samples (Improperly Cooked Pasta):
The remaining 1000 images represent improperly cooked pasta. These samples may show overcooked or burnt edges, raw or undercooked areas, uneven texture, sogginess, and discoloration. These images form the negative class and enable the model to identify defects and poor cooking conditions accurately.
Data Collection Setup
All images were captured using the Realme C25Y smartphone, equipped with a 50 MP rear camera system. The high-resolution camera ensured detailed image capture, making important visual features such as texture, color variation, and structural consistency clearly visible.
A simple, neutral background was used intentionally during image capture to enhance contrast between the pasta and the background and reduce distractions. Consistent lighting conditions were maintained throughout the data collection process to ensure accurate color representation and minimize environmental variations.
Image Characteristics
The dataset includes variations in pasta shape, size, and cooking style, such as boiled or baked preparations. It also captures different cooking defects, including burnt portions, raw areas, sogginess, and uneven consistency. This variation increases dataset diversity and improves the robustness of the classification model.
Data Annotation and Timeline
Each image is carefully labeled as either “Good” (Properly Cooked) or “Bad” (Improperly Cooked). These labels serve as ground truth for supervised learning. The images were collected over a period of seven days, during which pasta samples gradually transitioned from good to bad conditions. This approach helps the model learn realistic deterioration patterns, making it more reliable for practical applications.