Implementation:Open compass VLMEvalKit ConcatVideoDataset

Field	Value
source	VLMEvalKit
domain	Vision, Benchmarking, Video Dataset Aggregation

Overview

Benchmark dataset implementation for Concat Video evaluation in VLMEvalKit.

Description

ConcatVideoDataset inherits from VideoBaseDataset and provides a mechanism to aggregate multiple video datasets into a single evaluation dataset. It dynamically builds sub-datasets, concatenates their data with SUB_DATASET labels, and delegates evaluation to individual sub-dataset evaluate methods.

Usage

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

Code Reference

Source: vlmeval/dataset/video_concat_dataset.py, Lines: L1-85
Import: from vlmeval.dataset.video_concat_dataset import ConcatVideoDataset

Signature:

class ConcatVideoDataset(VideoBaseDataset):
    TYPE = 'dynamic'
    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('ConcatVideoDataset')

Related Pages

Page Connections

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

Principle

Implementation

Heuristic

Environment