Furniture Dataset for Multi-Angle AI Training
This dataset offers more than 10,000 furniture images captured from various angles to improve the accuracy of AI models for detection, classification, or segmentation. The objects are divided into 6 categories, with a large number of photos per class, and a variety of shots (light, perspective, background). Ideal for e-commerce, augmented reality and connected home projects.
Description
Furniture Dataset for Multi-Angle AI Training brings together over 10,000 annotated images of household furniture, captured from a variety of angles and conditions. It covers six main categories: chairs, coffee tables, dining tables, dressers, beds, bar stools. The aim is to support the robust detection and classification of objects in varied environments.
What is this dataset for?
- Form recognition models for furniture and decoration-oriented objects
- Improve applications in augmented reality or 3D design
- Powering visual recommendation systems in furniture e-commerce
Can it be enriched or improved?
Yes. It is possible to add new categories, to label materials, colors or even dimensions. The addition of segmentation masks or 3D data would also reinforce the value of the dataset for advanced uses.
🔎 In summary
🧠 Recommended for
- AR/VR app developers
- E-commerce startups
- AI vision researchers
🔧 Compatible tools
- YoloV5
- Detectron2
- Roboflow
- TensorFlow Object Detection API
💡 Tip
To detect objects in real conditions, use a data augmentation with a random background and a change in lighting.
Frequently Asked Questions
Does this dataset contain bounding boxes?
No, only class tags are available. It can be enriched manually or via annotation tools.
Can this dataset be integrated into a commercial augmented reality project?
Yes, the Apache 2.0 license allows free use, including commercial use.
What categories are available in this dataset?
The six categories are: chair, coffee table, dining table, dining table, dresser, bed, and bar stool. Each class includes several hundred images.
