yizhangdev/pascal-voc

Pascal VOC

Dataset Summary

The Pascal Visual Object Classes (VOC) dataset is a widely used benchmark in the field of computer vision. It is designed for object detection, image classification, semantic segmentation, and action classification tasks. The dataset provides a comprehensive set of annotated images covering 20 object classes, allowing researchers to evaluate and compare the performance of various algorithms. Note: This dataset repository contains all editions of PASCAL-VOC, each file is identified with the year.

Dataset Structure

Images: The dataset contains 178k images. Annotations: Annotations include object bounding boxes, object class labels, segmentation masks, and action labels. Classes: 20 object classes: person, bicycle, car, motorbike, aeroplane, bus, train, boat, bird, cat, dog, horse, sheep, cow, elephant, bear, zebra, giraffe, and potted plant. Supported Tasks Image Classification: Assigning a label to an image from a fixed set of categories. Object Detection: Identifying objects within an image and drawing bounding boxes around them. Semantic Segmentation: Assigning a class label to each pixel in the image. Action Classification: Identifying the action being performed in the image.

Applications

The Pascal VOC dataset is used for:

• —Benchmarking and evaluating computer vision algorithms.
• —Training models for image classification, object detection, and segmentation tasks.

Data Collection and Annotation

Data Sources The images were collected from Flickr and other sources, ensuring a diverse and representative sample of real-world scenes.

Annotation Process Annotations were carried out by a team of human annotators. Each image is labeled with:

• —Bounding boxes for object detection.
• —Class labels for each object.
• —Pixel-wise segmentation masks for semantic segmentation.
• —Action labels indicating the action performed by the objects in the image.

License

The Pascal VOC dataset is released under the Creative Commons Attribution 2.5 License. Users are free to share, adapt, and use the dataset, provided appropriate credit is given.

Citation

If you use the Pascal VOC dataset in your research, please cite the following paper:

@article{Everingham10,
  author    = {Mark Everingham and
               Luc Gool and
               Christopher K. I. Williams and
               John Winn and
               Andrew Zisserman},
  title     = {The Pascal Visual Object Classes (VOC) Challenge},
  journal   = {International Journal of Computer Vision},
  volume    = {88},
  number    = {2},
  year      = {2010},
  pages     = {303-338},
}