Review: The Open Images Dataset V4

Open Images Dataset, From V4 to V7

Sik-Ho Tsang
3 min readDec 18, 2022

The Open Images Dataset V4: Unified Image Classification, Object Detection, and Visual Relationship Detection at Scale
Open Images, by Google Research
2020 IJCV, Over 1400 Citations (Sik-Ho Tsang @ Medium)
Image Classification, Object Detection, Visual relationship Detection, Instance Segmentation, Dataset

  • In this paper, Open Images V4, is proposed, which is a dataset of 9.2M images with unified annotations for image classification, object detection and visual relationship detection.
  • And later on, the dataset is updated with V5 to V7:
  1. Open Images V5 features segmentation masks.
  2. Open Images V6 features localized narratives.
  3. Open Images V7 features point labels.

== Open Images V4 ==

Example annotations in Open Images for image classification, object detection, and visual relationship detection.
  • In V4, each image comes with labels of image classification, object detection, and visual relationship detection.

== Open Images V5 (2019) ==

Sample masks on the training set of Open Images V5.
  • Open Images V5 features segmentation masks for 2.8 million object instances in 350 categories, for instance segmentation.

== Open Images V6 (2020) ==

Sample of localized narratives
  • Open Images V6 provides localized narratives, which are generated by annotators who provide spoken descriptions of an image while they simultaneously move their mouse to hover over the regions they are describing.

== Open Images V7 (2022) ==

Example images with point-level labels. Circles ⭘ are “yes” labels, and squares are “no” labels.
  • In Open Image V7, for each annotated image, a collection of points is also provided, each with a “yes” or “no” label for a given class.
  • There is a total of 38.6M new point annotations (12.4M with “yes” labels) that cover 5.8 thousand classes and 1.4M images.

References

[2020 IJCV] [Open Images]
The Open Images Dataset V4: Unified Image Classification, Object Detection, and Visual Relationship Detection at Scale

[Google AI Blogs] [Open Images V5] [Open Images V6] [Open Images V7]
[Dataset] [Open Images V1 to V7]

1.1. Image Classification

1989 … 2020 [Open Images] … 2022 [ConvNeXt] [PVTv2] [ViT-G] [AS-MLP] [ResTv2] [CSWin Transformer] [Pale Transformer] [Sparse MLP] [MViTv2]

1.4. Object Detection

20142020 [Open Images] … 2022 [PVTv2] [YOLOv7] [Pix2Seq] [MViTv2]

1.6. Instance Segmentation

2014 … 2020 [Open Images] … 2021 [PVT, PVTv1] [Copy-Paste] 2022 [PVTv2]

==== My Other Previous Paper Readings ====

--

--

Sik-Ho Tsang

PhD, Researcher. I share what I learn. :) Linktree: https://linktr.ee/shtsang for Twitter, LinkedIn, etc.