RGB Image Dataset for Cabbage Maturity Classification at 100 and 135 Days After Transplanting (DAT) for Machine Learning Applications
Description
This dataset was developed to support research on automated crop maturity detection using computer vision and deep learning techniques. The hypothesis is that RGB image features of cabbage heads captured under field conditions contain sufficient information to distinguish between mature and premature harvesting stages. By training deep learning models, it is possible to predict the optimal harvesting stage, supporting precision agriculture and reducing subjectivity in decisions. The dataset consists of RGB images of cabbage plants collected from an experimental agricultural field. Images were captured at two growth stages defined by days after transplanting (DAT): 100 DAT (premature) and 135 DAT (mature). A total of 616 images were collected (320 mature cabbage samples (Class 1) and 296 premature samples (Class 2)). All images were captured using a 16-megapixel RGB camera (Canon PowerShot SX170) under natural field lighting conditions. The original resolution was 1632 × 1553 pixels, preserving key visual characteristics such as leaf arrangement, head compactness, and size. To ensure compatibility with deep learning architectures and reduce computational requirements, all images were resized to 224 × 224 pixels. This standardized input size enables efficient training of convolutional neural networks used in agricultural image analysis. Although resizing improves computational efficiency and standardizes model input, it may slightly reduce the ability to capture very fine details such as subtle leaf textures. However, the resized images retain the main morphological features needed for distinguishing maturity stages. The dataset shows clear visual differences between classes. Mature cabbages exhibit larger, denser, and more compact heads, while premature cabbages show looser leaf structures and less head formation. These patterns provide useful features for machine learning algorithms. This dataset supports applications including training and evaluation of deep learning models for crop maturity detection, development of computer vision systems for automated harvesting, research in precision agriculture and smart farming, and benchmarking algorithms for agricultural image classification. Each image is labeled according to its maturity stage: Class 1 (mature, 135 DAT) and Class 2 (premature, 100 DAT), enabling easy integration with machine learning workflows. Overall, this dataset provides a valuable resource for advancing research in AI-based crop maturity prediction and intelligent agricultural harvesting systems. The authors gratefully acknowledge financial support from ICAR–CIAE under the CRP-FMPF project.
Files
Steps to reproduce
This dataset was developed to support research on automated crop maturity detection using computer vision and deep learning techniques. The hypothesis is that RGB image features of cabbage heads captured under field conditions contain sufficient information to distinguish between mature and premature harvesting stages. By training deep learning models, it is possible to predict the optimal harvesting stage, supporting precision agriculture and reducing subjectivity in decisions. The dataset consists of RGB images of cabbage plants collected from an experimental agricultural field. Images were captured at two growth stages defined by days after transplanting (DAT): 100 DAT (premature) and 135 DAT (mature). A total of 616 images were collected (320 mature cabbage samples (Class 1) and 296 premature samples (Class 2)). All images were captured using a 16-megapixel RGB camera (Canon PowerShot SX170) under natural field lighting conditions. The original resolution was 1632 × 1553 pixels, preserving key visual characteristics such as leaf arrangement, head compactness, and size. To ensure compatibility with deep learning architectures and reduce computational requirements, all images were resized to 224 × 224 pixels. This standardized input size enables efficient training of convolutional neural networks used in agricultural image analysis. Although resizing improves computational efficiency and standardizes model input, it may slightly reduce the ability to capture very fine details such as subtle leaf textures. However, the resized images retain the main morphological features needed for distinguishing maturity stages. The dataset shows clear visual differences between classes. Mature cabbages exhibit larger, denser, and more compact heads, while premature cabbages show looser leaf structures and less head formation. These patterns provide useful features for machine learning algorithms. This dataset supports applications including training and evaluation of deep learning models for crop maturity detection, development of computer vision systems for automated harvesting, research in precision agriculture and smart farming, and benchmarking algorithms for agricultural image classification. Each image is labeled according to its maturity stage: Class 1 (mature, 135 DAT) and Class 2 (premature, 100 DAT), enabling easy integration with machine learning workflows. Overall, this dataset provides a valuable resource for advancing research in AI-based crop maturity prediction and intelligent agricultural harvesting systems. The authors gratefully acknowledge financial support from ICAR–CIAE under the CRP-FMPF project.