HyperAIHyperAI

Command Palette

Search for a command to run...

Image Understanding Benchmark Dataset

Date

Organization

Microsoft Corporation

License

cdla-permissive-2.0

The Image Understanding Benchmark is a synthetic dataset released by Microsoft in 2024 for image understanding evaluation, designed to assess the capabilities of multimodal models in areas such as image understanding, spatial reasoning, visual prompting, object recognition, and detection.

The dataset comprises four sub-tasks: object detection, object recognition, spatial reasoning, and visual prompting, generating approximately 10,240 images in total. The data is generated programmatically by combining COCO objects with Places365 backgrounds, incorporating random rotations, positional shifts, and scaling variations, and can be used to evaluate the visual understanding capabilities of multimodal models.

Dataset Composition

The dataset includes 8 configurations (configs), divided into single-object (single) and two-object (pairs) conditions, covering the following four sub-tasks:

  • Object Detection
  • Object Recognition
  • Spatial Reasoning
  • Visual Prompting

Each sub-task generates 1,280 images under each of the two conditions, totaling approximately 10,240 images. Each configuration primarily contains the following fields:

  • id: Unique identifier for the data sample.
  • image: Input image.
  • prompt: Prompt text used for model evaluation.
  • ground_truth: Ground truth for certain tasks.
数据集示例
数据集示例

Build AI with AI

From idea to launch — accelerate your AI development with free AI co-coding, out-of-the-box environment and best price of GPUs.

AI Co-coding
Ready-to-use GPUs
Best Pricing

HyperAI Newsletters

Subscribe to our latest updates
We will deliver the latest updates of the week to your inbox at nine o'clock every Monday morning
Powered by MailChimp