Implementation:Open compass VLMEvalKit MIABench

Field	Value
source	VLMEvalKit
domain	Vision, Benchmarking, Instruction Adherence

Overview

Benchmark dataset implementation for MIA-Bench evaluation in VLMEvalKit.

Description

MIABench inherits from ImageBaseDataset and implements the MIA-Bench benchmark for evaluating multimodal instruction adherence. The TYPE field is set to 'VQA'. It uses GPT-4o as default judge for evaluating instruction following quality.

Usage

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

Code Reference

Source: vlmeval/dataset/miabench.py, Lines: L1-166
Import: from vlmeval.dataset.miabench import MIABench

Signature:

class MIABench(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('MIA-Bench')

Related Pages

Page Connections

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

Principle

Implementation

Heuristic

Environment