Implementation:Open compass VLMEvalKit WildVision

Field	Value
source	VLMEvalKit
domain	Vision, Benchmarking, In-the-wild Vision

Overview

Benchmark dataset implementation for WildVision evaluation in VLMEvalKit.

Description

WildVision inherits from ImageBaseDataset and implements the WildVision benchmark for evaluating models on in-the-wild vision tasks. The TYPE field is set to 'VQA'. It uses a pairwise comparison scoring system (A>>B through B>>A) for relative model evaluation.

Usage

Registered in vlmeval/dataset/__init__.py and invoked through build_dataset() by benchmark name.

Code Reference

Source: vlmeval/dataset/wildvision.py, Lines: L1-222
Import: from vlmeval.dataset.wildvision import WildVision

Signature:

class WildVision(ImageBaseDataset):
    TYPE = 'VQA'
    DATASET_URL = {...}
    DATASET_MD5 = {...}
    ...

I/O Contract

Direction	Description
Inputs	TSV dataset file with image/video paths and questions
Outputs	Evaluation results DataFrame with scores per category

Usage Examples

from vlmeval.dataset import build_dataset
dataset = build_dataset('WildVision')

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment