Mustafa5645344/specifly-insect-dataset

Specifly: 50K Augmented Agricultural Insect Dataset

This dataset contains 49,680 augmented RGB images of agricultural pests and beneficial insects across 21 biological taxa, generated and annotated via Roboflow.

Dataset Overview

• Total Images: 49,680
• Primary Task: Multi-class Insect Detection (YOLOv8 bounding boxes) & Fine-Grained Classification
• Domain: Agricultural crop protection, entomology, and autonomous farm robotics

Species / Classes Covered (21 Taxa)

1. ant
2. aphid
3. bees
4. butterfly
5. caterpillar
6. cicada
7. dragonfly
8. grasshopper
9. green_lacewing
10. ladybug
11. leafhopper
12. mantis
13. mole_cricket
14. planthopper
15. rhino_beetle
16. rice_bug
17. spider
18. stem_borer
19. stink_bug
20. undefined
21. weevil

How to Access & Download Dataset

Option 1: Direct Roboflow SDK Integration

from roboflow import Roboflow

rf = Roboflow(api_key="YOUR_ROBOFLOW_KEY")
project = rf.workspace("specifly").project("specifly-3-rwm7i")
version = project.version(1)
dataset = version.download("yolov8")

Option 2: Pre-trained Models

You can utilize the models trained on this dataset directly from Hugging Face Hub:

• Vision Transformer (ViT) Classifier: Mustafa5645344/insect-detection-vit
• YOLOv8 Detector: Mustafa5645344/insect-detection-yolov8