Skip to content

Latest commit

 

History

6 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 

Repository files navigation

FruitVision-Dataset

Version: v1.0

  • Last Updated: April 2025
  • License: MIT License

Overview

The FruitVision Dataset is a high-quality, expert-validated image dataset designed for the classification of fruits under different physical states: fresh, rotten, and formalin-mixed (toxic). It is intended for use in computer vision research, machine learning model development, and agricultural health informatics applications. The dataset includes images of Apple, Banana, Mango, Orange, and Grapes captured in real-world conditions using different smartphone devices and lighting setups. It also provides a wide variety of augmentations and preprocessing for better generalization in model training.

Metadata

Attribute Description
Dataset Name FruitVision Dataset
Version v1.0
Last Updated April 2025
Fruits Included Apple, Banana, Mango, Orange, Grapes
Categories Fresh, Rotten, Formalin-Mixed (Toxic)
Total Images [Original: 10,154 & Augmented: 81,232]
File Format .jpg
Resolution 512 × 512 pixels
Capture Devices iPhone 15 Pro Max, Redmi POCO M2 Reloaded, Redmi Note 9 Pro
Capture Conditions Daylight, Indoor lighting, Plain backgrounds
Annotation Manual, verified by agricultural domain experts
Augmentation Rotation, Flip, Zoom, Brightness, Shear, Gaussian Noise
Preprocessing Resizing to 512×512 using OpenCV
Intended Use Fruit classification, spoilage detection, food safety ML

Preprocessing and Augmentation

  • Augmentation performed using imgaug and OpenCV libraries includes:
  1. Rotation: 45°, 60°, 90°
  2. Flip: Horizontal Flip (50% chance)
  3. Zooming: Scale range 0.8–1.2
  4. Brightness: Multiplication by 0.8–1.2
  5. Shearing: Range from -16 to +16 degrees
  6. Noise: Additive Gaussian Noise

Download dataset:

Citation:

Bijoy, Md Hasan Imam; Tasnim, Syeda Zarin; Awsaf, Syed Ali; Hasan, Md Zahid (2025), “FruitVision: A Benchmark Dataset for Fresh, Rotten, and Formalin-mixed Fruit Detection”, Mendeley Data, V2, doi: 10.17632/xkbjx8959c.2

Ethical Statement:

All procedures for the fruit data collection were conducted in compliance with Daffodil International University's (DIU) ethical guidelines and relevant regulations. Ethical approval was granted by the Research Ethics Committee, Faculty of Science and Information Technology, DIU, under the approval number REC-FSIT-2024-04-17. This approval followed a comprehensive review process, ensuring adherence to safety protocols and ethical standards. The formalin-mixed fruits were strictly used for research purposes, validated by agricultural experts, and were not intended for consumption or sale to consumers

Acknowledgement

We extend our heartfelt gratitude to Professor Dr. M. A. Rahim, Head of the Department of Agricultural Science at Daffodil International University (DIU), Dhaka, Bangladesh, for his invaluable expertise in data validation. His insightful feedback and unwavering support were instrumental in the successful completion of this project.

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

1 watching

Forks

Releases

Packages

Contributors

Languages