Research map / Vision-language and multimodal / Multimodal LLMs
See the Glass Half Full: Reasoning About Liquid Containers, Their Volume and Content
Roozbeh Mottaghi, Connor Schenck, Dieter Fox, Ali Farhadi
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 |
|---|---|---|---|
| add context tensor from object detector masks (ablation) | CRC · COQE volume estimation | avg per-class accuracy: 15.33 -> 17.79 | Improved |
| add context tensor from object detector masks (ablation) | CRC · COQE content estimation | avg per-class accuracy: 29.29 -> 32.01 | Improved |
| add context tensor from object detector masks (ablation) | CRC Siamese · COQE comparative volume estimation | avg per-class accuracy: 48.97 -> 49.81 | Improved |
| add context tensor from object detector masks (ablation) | CRC+RNN · COQE pouring prediction (exact match) | sequence accuracy: 28.97 -> 30.13 | Improved |
| add context tensor from object detector masks (ablation) | CRC+RNN · COQE pouring prediction (edit distance 2-3) | sequence accuracy: 40.03 -> 39.96 | No clear effect |
| use ResNet-18 with context instead of AlexNet | CRC · COQE pouring prediction (exact match) | sequence accuracy: 21.39 -> 30.13 | Improved |
| use ResNet-18 CRC with context instead of AlexNet | CRC · COQE comparative volume estimation | avg per-class accuracy: 43.90 -> 49.81 | Improved |
| use ResNet-18 CRC without context instead of AlexNet | ResNet-18 · COQE volume estimation | avg per-class accuracy: 17.33 -> 15.33 | Hurt |
| use classification instead of regression (ablation) | volume estimation · COQE volume estimation | accuracy | Improved |
| share weights across the Siamese branches (ablation) | comparative volume network · COQE comparative volume estimation | accuracy | Improved |
| use two-branch Siamese input instead of 5-channel input with both masks (ablation) | comparative volume network · COQE comparative volume estimation | accuracy | Improved |
Closest papers by meaning
- PourIt!: Weakly-Supervised Liquid Perception from a Single Image for Visual Closed-Loop Robotic Pouring (ICCV 2023 · similarity 0.87)
- Learning to See Inside Opaque Liquid Containers using Speckle Vibrometry (ICCV 2025 · similarity 0.87)
- The Sound of Water: Inferring Physical Properties From Pouring Liquids (TPAMI 2026 · similarity 0.87)
- Physical Property Understanding from Language-Embedded Feature Fields (CVPR 2024 · similarity 0.80)
- Leveraging RGB-D Data with Cross-Modal Context Mining for Glass Surface Detection (AAAI 2025 · similarity 0.80)
- Objects as Volumes: A Stochastic Geometry View of Opaque Solids (CVPR 2024 · similarity 0.80)
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.
