[KongchangAI]
BenchmarkD-Vlog 数据集

D-Vlog

A video blog dataset for multimodal depression detection that contains multimodal data such as speech, visual and language modalities. It is commonly used to evaluate depression risk prediction models.

Timeline (last 90 days)

Nov 18

With a frozen 7B backbone, EDA Support improves F1 by 7.61 and 9.88 percentage points over direct prediction on D-Vlog Leave-Query-Out and LMVD, respectively.

Unverified50%
Nov 18

D-Vlog and LMVD are two commonly used multimodal depression detection datasets; D-Vlog is based on video blog self-narratives and contains segment-level emotion and depression-related labels; LMVD targets interview-style scenarios and contains structured question-and-answer dialogue and scale scores.

Unverified50%
Nov 13

Multimodal fusion reaches 98.41% F1 on the D-Vlog dataset.

Unverified50%

All Facts (3)

Source Articles