DOSCH AI-Datasets: Agricultural Tractor

2D DATASET: Synthetic Data for Computer Vision.

Accelerate the development of your smart farming applications with high-quality synthetic training data. This dataset contains 3,240 pre-rendered views of an agricultural tractor, including pixel-perfect ground truth data for depth, segmentation, and surface structure.

Ideal for training neural networks in the fields of autonomous driving, object detection, and safety surveillance in agriculture.

Technical Specifications

• Total Images: 3,240 ((Number of images: 648 x 5 environment variants (Total: 3,240))
• Resolution: 2048 x 2048 px (High-Res)
• Variations: 5 Lighting Variations
• Formats: EXR (32-bit Float) + PNG (Alpha)

Why DOSCH AI-Datasets?

• Ready-to-Train: No 3D software or rendering pipeline needed.
• Perfect Ground Truth: Pixel-accurate data, free from human annotation errors.
• Lighting Variations: 5 different lighting moods per view for robust AI training.
• Data Augmentation Friendly: Can be mounted on any scene thanks to isolated backgrounds (Alpha channel).
• Legally Safe & Royalty Free: 100% synthetically generated. No GDPR/privacy issues, clear copyright situation.
  
   

Included Data Channels

You receive four perfectly synchronized files in professional industry formats for each of the 3,240 images:

1. RGB Color

Format: PNG (24-bit + Alpha)
Photorealistic rendering of the tractor with transparent background.
Benefit: Enables easy "Cut-and-Paste" augmentation onto your own background images.

2. Semantic Segmentation

Format: PNG (Hard Edge)
Precise color mask ("Clown Mask") representing the exact silhouette of the tractor.
Benefit: Automatic classification and pixel-perfect separation from the background.

3. Depth Map / Z-Depth

Format: EXR (32-bit Float)
Contains the exact distance information from the camera sensor to the object for every pixel.
Benefit: Essential for depth estimation, distance measurement, and robotics.

4. Normal Map

Format: EXR (32-bit Float)
Depicts the surface orientation (normals) of the geometry in space.
Benefit: Helps the AI understand 3D shape and structure independent of texture and color.

Use Cases

• Autonomous Farming: Training vision systems for self-driving tractors.
• Fleet Management: Detection of vehicle types via drone.
• Safety & Surveillance: Human detection vs. machine detection (Safety Zones).
• Synthetic Data Gen: Use in generators for synthetic environments.

Compatible with all common ML frameworks (PyTorch, TensorFlow, Keras, YOLO).