PaperClip Dataset

Key Features:

1. Paper Clip Sizes:
   • Small: 24×7 mm
   • Medium: 32×9 mm
   • Large: 44×11 mm
2. Defect Types:
   • Bending
   • Twisting
   • Asymmetrical warping
3. Dataset Composition:
   • Total Clips: 247 (127 normal, 120 defective)
   • Training Set: 68 images with 110 normal clips and 107 defective clips
   • Validation Set: 12 images with 17 normal and 17 defective clips (~15% of the dataset)
   • Test Video: A 23-second .mp4 video (~690 frames), showcasing additional defects not included in the training and validation sets, for real-world evaluation.

Dataset Highlights:

• High Resolution: Images are captured in 1920×1080, ensuring clarity and precision for analysis.
• Balanced Distribution: Ensures an equitable representation of normal and defective clips for effective model training.
• Variability: Includes a range of clip orientations, positions, and defect types to mimic real-world scenarios.

Applications:

This dataset is ideal for:

• Defect detection models
• Computer vision projects
• Object classification tasks
• Real-time evaluation using video data

Dataset Structure:

Test Video

The provided 23-second test video offers additional complexity by including unseen defect types, enabling developers to assess model performance on new scenarios.