Research map / Vision-language and multimodal / Visual question answering
Divide and Conquer: Answering Questions With Object Factorization and Compositional Reasoning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| add object factorization and compositional reasoning to XNM | XNM · VQA Novel | VQA accuracy: 52.81 -> 54.82 | Helped |
| add object factorization and compositional reasoning to XNM | XNM · GQA Novel | accuracy: 57.54 -> 59.71 | Helped |
| add object factorization and compositional reasoning to XNM | XNM · Novel-VQA Novel-Q | VQA accuracy: 57.54 -> 60.73 | Helped |
| add object factorization and compositional reasoning to XNM | XNM · VQA-CP | accuracy: 51.54 -> 53.99 | Helped |
| add object factorization and compositional reasoning to XNM | XNM · GQA-OOD | Δ: 14.4 -> 10.7 | Helped |
| replace factorized prototypes with randomly initialized prototypes (ablation) | POEM · VQA-CP OOD | accuracy: 53.99 -> 50.39 | Hurt |
| replace factorized prototypes with manually defined object prototypes (ablation) | POEM · VQA-CP OOD | accuracy: 53.99 -> 52.23 | Hurt |
| replace factorized prototypes with Glove concept embeddings (ablation) | POEM · VQA-CP OOD | accuracy: 53.99 -> 51.09 | Hurt |
| use factorized prototypes instead of manually defined object prototypes (ablation) | POEM · Novel-VQA Novel-Q | VQA accuracy: 56.67 -> 60.73 | Helped |
| use prototypes within the Find, Filter, and Describe modules rather than only in an independent Prototype module (ablation) | POEM · GQA Known | accuracy: 55.01 -> 60.60 | Helped |
| add the semantic memory module (ablation) | POEM · GQA Known | accuracy: 58.39 -> 60.60 | Helped |
| add the semantic memory module (ablation) | POEM · Novel-VQA Novel-Q | VQA accuracy: 59.41 -> 60.73 | Helped |
Closest papers by meaning
- Exploring the Effectiveness of Object-Centric Representations in Visual Question Answering: Comparative Insights with Foundation Models (ICLR 2025 · similarity 0.90)
- Composition-Grounded Instruction Synthesis for Visual Reasoning (ICLR 2026 · similarity 0.89)
- Visual Graph Arena: Evaluating Visual Conceptualization of Vision and Multimodal Large Language Models (ICML 2025 · similarity 0.88)
- ConMe: Rethinking Evaluation of Compositional Reasoning for Modern VLMs (NeurIPS 2024 · similarity 0.87)
- Core-to-Global Reasoning for Compositional Visual Question Answering (AAAI 2025 · similarity 0.87)
- Detection-Based Intermediate Supervision for Visual Question Answering (AAAI 2024 · similarity 0.87)
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.