Back to datasets
Dataset assetOpen Source CommunityImage RecognitionMaterial Images
UJM TIV
UJM TIV is a material dataset that contains images of 11 material categories, each image being a 20 × 20 RGB picture. The dataset aims to provide images with higher intra‑class variability, captured from four different viewpoints and four lighting conditions.
Source
github
Created
Oct 9, 2022
Updated
Dec 9, 2022
Signals
103 views
Availability
Linked source ready
Overview
Dataset description and usage context
UJM TIV Dataset Overview
Dataset Content
- Number of Classes: 11 material categories, including aluminium foil, brown bread, corduroy, cork, cotton, biscuits, lettuce leaf, linen, white bread, wood, and wool.
- Image Format: Each image is a 20 × 20 pixel RGB image.
Dataset Characteristics
- High Intra‑Class Variability: Provides images with substantial within‑class variation.
- Acquisition Conditions: Images were captured under controlled settings using four different viewpoints (front/90°, 10°, 30°, 60°) and four lighting conditions (front/90°, 20°, 45°, 65°) together with two sample orientations.
Intended Applications
- Research Purpose: Supports multi‑view learning solutions for material classification, employing a dual‑branch Siamese network that extracts and fuses information from two different viewpoints, achieving performance superior to conventional single‑view deep learning approaches.
Citation
- Reference: When using the UJM TIV dataset in scientific publications, please cite the following paper:
Contact
- Inquiry Email: borhancse.cu@gmail.com
Need downstream help?
Pair the dataset with AI analysis and content workflows.
Once the source passes your review, move straight into summarization, transformation, report drafting, or presentation generation with the JuheAI toolchain.