Datasets for image classification and detection

last modified : 19-07-2019

Here are listed all the datasets that can be used for image classification. Each dataset should come with a small description of its size, what's in it and who provided it.

The UA-DETRAC Benchmark Suite

This dataset is both for multi-object detection and multi-object tracking.

  • Contains:

    • Vehicles: Cars, Bus, Van, Other
    • Weather: cloudy, sunny, rainy, night
    • Different level of occlusion
  • Size:

    • more than 140 thousand frames
    • 8250 vehicles manually annotated
    • 1.21 million labeled bounding boxes of objects
  • Other details:

    • location: 24 different locations at Beijing and Tianjin in China
    • 10 hours of videos captured with a Cannon EOS 550D camera
  • Article:

    • Title: UA-DETRAC: A New Benchmark and Protocol for Multi-Object Detection and Tracking
    • Authors: Longyin Wen, Dawei Du, Zhaowei Cai, Zhen Lei, Ming{-}Ching Chang, Honggang Qi, Jongwoo Lim, Ming{-}Hsuan Yang and Siwei Lyu
    • Link: article
  • Dataset: here

The KITTI Vision Benchmark Suite

This dataset is both for multi-object detection and multi-object tracking.

  • Contains:

    • Tracking: 8 classes but only 'Car' and 'Pedestrian' have enough instance according to the website
    • Detection / 2D Objects: Unspecified on website, cars at least
    • Detection / 3D Objects: Unspecified on website, cars at least
    • Detection / bird's eye view: Cars (maybe)
  • Size:

    • Tracking: 21 training sequences and 29 test sequences
    • Detection / 2D Objects: 7481 training images and 7518 test images
    • Detection / 3D Objects: 7481 training images and 7518 test images
    • Detection / bird's eye view: 7481 training images and 7518 test images for 80 256 labeled objects
  • Article:

    • Title: Are we ready for Autonomous Driving? The KITTI Vision Benchmark Suite
    • Authors: Andreas Geiger and Philip Lenz and Raquel Urtasun
    • Link: article
  • Dataset: here

The Berkeley BBD100K

  • Contains:

    • Vehicles: Bus, Light, Sign, Person, Bike, Truck, Motor, Car, Train, Rider
    • Weather: clear, partly cloudy, over-cast, rainy, snowy, foggy, dawn/dusk, daytime, night
    • Different level of occlusion
    • Segmentation
    • Different scenes, such as: residential, highway, city, street, ...
    • Lane marking
  • Size:

    • 100,000 HD video sequences of over 1,100-hour driving experience;
    • 2D Bounding Boxes annotated on 100,000 images;
    • Segmentation over 10,000 diverse images with pixel-level and rich instance-level annotations;
    • Multiple types of lane marking annotations on 100,000 images.
  • Other details:

    • location: Different location in the USA, New York, Berkeley, San Francisco
  • Article:

    • Title: BDD100K: A Diverse Driving Video Database withScalable Annotation Tooling
    • Authors: Fisher Yu, Wenqi Xian, Yingying Chen, Fangchen Liu, Mike Liao, Vashisht Madhavan, Trevor Darrell
    • Link: article
  • Dataset: here

Cifar-10

  • Contains:

    • airplane
    • automobile
    • bird
    • cat
    • deer
    • dog
    • frog
    • horse
    • ship
    • truck
  • Size:

    • 60000 images divided into 6 batches with one for the tests
    • images of size 32x32
  • Dataset: here

VKitti

  • Contains:

    • Based on the Kitti dataset, should contain the same classes
  • Size:

    • 50 high-resolution monocular videos (21,260 frames)
  • Other details:

    • five different virtual worlds in urban settings under different imaging and weather conditions
    • These photo-realistic synthetic videos are automatically, exactly, and fully annotated for 2D and 3D multi-object tracking and at the pixel level with category, instance, flow, and depth labels
  • Article:

    • Authors: Gaidon A, Wang Q, Cabon Y and Vig E
    • Link: article
  • Dataset: here