rf100-vl-datasets / deepfruits /README.dataset.txt
gatilin's picture
Upload folder using huggingface_hub
33e8cfd verified
# deepfruits-mango-xs1as-tkygw-uxnm > 2025-03-13 3:17pm
https://universe.roboflow.com/rf-100-vl/deepfruits-mango-xs1as-tkygw-uxnm
Provided by a Roboflow user
License: MIT
# Overview
- [Introduction](#introduction)
- [Object Classes](#object-classes)
- [Class 1](#class-1)
# Introduction
The dataset focuses on detecting mangoes on trees, aiding in agricultural management and harvest optimization. The dataset consists of images captured in various lighting and occlusion conditions, with a single class:
- Mango: Mango fruits hanging on trees.
# Object Classes
## Class 1: Mango
### Description
Mangoes are oval or oblong-shaped fruits that can vary in size. They typically appear smooth and are often found attached to tree branches. Their surface can show gradients when captured in sunlight.
### Instructions
- **Annotate all visible mangoes**: Draw a bounding box around each mango in the image, ensuring to cover the entire visible area of the fruit.
- **Handle occlusions carefully**: If a mango is partially obscured by leaves or branches, guess the occluded portion based on visible parts.
- **Edge scenarios**: If a mango is cut off by the image boundary, stop the bounding box at the edge.
- **Avoid small or unclear visuals**: Do not annotate objects that are too small, don't too confidently identify visually unclear objects as mangoes.
- **Exclude non-mango objects**: Ensure you are not labeling leaves, branches or other tree parts; focus solely on the mangoes.