AI Paper Atlas

Research map / Vision-language and multimodal / Visual question answering

Question-Guided Hybrid Convolution for Visual Question Answering

Peng Gao, Hongsheng Li, Shuang Li, Pan Lu, Yikang Li, Steven C.H. Hoi, Xiaogang Wang

ECCV 2018 · Vision-language and multimodal › Visual question answering · approach: Attention and prior-robust VQA · 79 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add group shuffling layer (ablation)QGHC · VQA valaccuracy (All): 58.92 -> 59.24Improved
use 8 groups instead of 4 (ablation)QGHC · VQA valaccuracy (All): 59.01 -> 59.24Improved
use 8 groups instead of 16 (ablation)QGHC · VQA valaccuracy (All): 58.22 -> 59.24Improved
stack 3 modules instead of 1 (ablation)QGHC · VQA valaccuracy (All): 58.88 -> 59.24Improved
stack 4 modules instead of 3 (ablation)QGHC · VQA valaccuracy (All): 59.24 -> 59.13No clear effect
double output channels (vs halved) (ablation)QGHC · VQA valaccuracy (All): 58.78 -> 59.24Improved
add residual 1x1-3x3-1x1 structure to naive full convolution (ablation)QGHC-1-naive · VQA valaccuracy (All): 55.32 -> 57.01Improved
convert full convolutions to 8-group convolutions (ablation)QGHC-1-full · VQA valaccuracy (All): 57.01 -> 58.41Improved
use hybrid (dynamic plus freely-updated) kernels instead of all-predicted group kernels (ablation)QGHC-1-group · VQA valaccuracy (All): 58.41 -> 58.88Improved
concatenate QGHC features with question features (ablation)QGHC · VQA valaccuracy (All): 59.24 -> 59.80Improved
fuse QGHC features with question features via MUTAN (ablation)QGHC · VQA valaccuracy (All): 59.24 -> 60.13Improved
add attention (ablation)QGHC · VQA valaccuracy (All): 59.24 -> 60.64Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy