Dataset Viewer

The dataset viewer is not available because its heuristics could not detect any supported data files. You can try uploading some data files, or configuring the data files location manually.

EdgePCB Defect Detection Dataset

Overview

The EdgePCB Defect Detection Dataset is a large-scale annotated dataset designed for training and evaluating deep learning models for automated PCB (Printed Circuit Board) inspection. The dataset supports real-time object detection tasks and is specifically curated for industrial applications using edge AI systems.

It is developed as part of the EdgePCB-AI project, which focuses on deploying YOLOv8-based defect detection models on NVIDIA Jetson Orin for real-time inference in manufacturing environments.


Dataset Details

  • Total Images: ~28,000

  • Task: Object Detection

  • Annotation Format: YOLO (bounding boxes in .txt format)

  • Image Resolution: 640×640 / 512×512 (resized)

  • Dataset Split:

    • Train: ~84%
    • Validation: ~10%
    • Test: ~6%

Classes

The dataset includes annotations for the following 6 PCB defect classes:

  1. Missing Hole
  2. Mouse Bite
  3. Open Circuit
  4. Short Circuit
  5. Spur
  6. Spurious Copper

Data Collection and Annotation

The dataset is compiled from publicly available PCB defect datasets and further processed for consistency and usability. All images are annotated in YOLO format using tools such as Roboflow and LabelImg.

Data preprocessing includes:

  • Image resizing
  • Normalization
  • Data augmentation (flip, rotation, brightness variation, mosaic augmentation)

Use Cases

This dataset is suitable for:

  • Real-time PCB defect detection
  • Industrial quality inspection systems
  • Edge AI deployment (Jetson Nano / Xavier / Orin)
  • Training YOLO, SSD, Faster R-CNN, and other object detection models
  • Research in computer vision for manufacturing automation

Benchmark and Performance

This dataset has been used to train YOLOv8 models achieving high detection accuracy (mAP@0.5 > 0.9) with real-time inference performance (~25–30 FPS) on NVIDIA Jetson Orin using TensorRT optimization.


License

MIT


Acknowledgements

This dataset is developed as part of an academic project at Vishwakarma Institute of Information Technology, Pune, and is inspired by publicly available PCB inspection datasets.


Citation

If you use this dataset in your research, please cite:

EdgePCB-AI: Real-Time PCB Defect Detection Using Embedded Vision

Downloads last month
32