Research map / Vision-language and multimodal / Visual question answering
Differential Networks for Visual Question Answering
Chenfei Wu, Jinlai Liu, Xiaojie Wang, Ruifan Li
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 changed | Where | Measured | Effect |
|---|---|---|---|
| replace fusion between feature elements (FCN-based, Mutan) with fusion of pairwise differences (DN) (ablation) | DF fusion · VQA 2.0 validation | accuracy: 63.61 -> 64.89 | Improved |
| use DN on image features in addition to questions (ablation) | DF · VQA 2.0 validation | accuracy: 64.46 -> 64.89 | Improved |
| use DN on question features in addition to images (ablation) | DF · VQA 2.0 validation | accuracy: 64.58 -> 64.89 | Improved |
| add dropout (ablation) | DF · VQA 2.0 validation | accuracy: 61.05 -> 64.89 | Improved |
| use relu instead of tanh (ablation) | DF · VQA 2.0 validation | accuracy: 64.78 -> 64.89 | No clear effect |
| DF vs prior state of the art on VQA 1.0, COCO-QA, TDIUC | VQA models · COCO-QA test | accuracy (All): 66.49 -> 69.36 | Improved |
| DF (36 boxes) vs Up-Down (36 boxes) | VQA 2.0 · VQA 2.0 test-dev | accuracy (All): 65.32 -> 67.73 | Improved |
| increase DF image features from 36 to 100 boxes | DF · VQA 2.0 test-std | accuracy (All): 67.86 -> 68.59 | Improved |
| DF vs LC Counting (100 boxes) on VQA 2.0 test-dev | VQA 2.0 · VQA 2.0 test-dev | accuracy (All): 68.09 -> 68.31 | Mixed |
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.
- Mutan(5) — MUTAN: Multimodal Tucker Fusion for Visual Question Answering (ICCV 2017) · better on 2 of 2 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
- Don't Just Assume; Look and Answer: Overcoming Priors for Visual Question Answering (CVPR 2018 · similarity 0.92)
- An Analysis of Visual Question Answering Algorithms (ICCV 2017 · similarity 0.92)
- Making the v in VQA Matter: Elevating the Role of Image Understanding in Visual Question Answering (CVPR 2017 · similarity 0.91)
- When Open-Vocabulary Visual Question Answering Meets Causal Adapter: Benchmark and Approach (AAAI 2025 · similarity 0.90)
- Object Attribute Matters in Visual Question Answering (AAAI 2024 · similarity 0.90)
- FRAMES-VQA: Benchmarking Fine-Tuning Robustness across Multi-Modal Shifts in Visual Question Answering (CVPR 2025 · similarity 0.90)
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.
