Hand-Object Interaction & Affordance
Fine-Grained 3D Hand Affordance Grounding and Generation for Small-Part Articulated Interactions
Generated automatically from the limitations stated in 4 papers (CVPR, ICLR, ICCV), listed under Evidence. It is not a paper, and it does not come from papers submitted to CSPaper.
The problem
Current affordance learning pipelines rely on self-training loops, video extraction, or upstream generative HOI models that fail when contact targets are small functional components such as bottle caps, zippers, or rotary knobs. Because generated hand poses easily drift to adjacent object surfaces, models exhibit catastrophic failure on precision affordances (e.g., 0% accuracy on pull or sub-50% on twist), propagating noisy pseudo-labels across unobserved geometries. Downstream manipulation systems are consequently blocked from executing precision tasks like uncapping, pulling sliders, or fine bimanual manipulation.
Why it matters
Enables robotic and generative systems to accurately localize and synthesize viable hand interactions on millimeter-to-centimeter scale functional object parts, unlocking precision manipulation tasks such as unscrewing caps, unzipping bags, and operating small switches.
Ways to approach it
Prior-work checks are free with an account. Results someone already ran are shown to everyone.
- 1
Contact-constrained local refinement: Develop a local geometry-aware refinement module that conditions hand pose and affordance heatmaps on high-resolution local surface curvature and keypoint signifiers rather than global object representations, evaluated by contact intersection volume and part-level affordance precision on AffordPose and SceneFun3D benchmarks.
- 2
Signifier-guided pseudo-label filtering: Build a geometric consistency verifier that filters noisy pseudo-labels in self-training loops using functional affordance priors (e.g., surface normals, clearance margins), measured by label error rate and generalization accuracy on out-of-distribution small-part test shapes.
- 3
Multi-view fine-grained interaction bootstrapping: Synthesize precision contact regions by projecting fine-scale multi-view contact trajectories into 3D signed distance fields, measured by affordance mIoU and downstream grasp success for precision affordance classes (twist, pull, pinch).
Have a different approach?
Describe how you would tackle this problem and we'll look for papers that already do it. Free; your text stays private.
Why it might fail
Sensor noise and low point cloud resolution in real-world 3D scans might fundamentally obscure millimeter-scale parts like zipper tabs, rendering purely geometric refinement insufficient without specialized macro-sensing.
Evidence
Each paper's own statement of the limitation, verbatim.
- H2OFlow: Grounding Human-Object Affordances with 3D Generative Models and Dense Diffused FlowsICLR 2026
Quality of affordance generation depends on the coverage and fidelity of synthetic HOI data generated by the upstream 3D generative model (CHOIS).
- AffordGrasp: Cross-Modal Diffusion for Affordance-Aware Grasp SynthesisCVPR 2026
Affordance generation relies on pseudo-labels generated by a self-training loop, which may propagate annotation noise on highly out-of-distribution geometries.
- VidBot: Learning Generalizable 3D Actions from In-the-Wild 2D Human Videos for Zero-Shot Robotic ManipulationCVPR 2025
Precise tasks like unscrewing caps remain challenging with the current extracted affordance data
- AffordPose: A Large-Scale Dataset of Hand-Object Interactions with Affordance-Driven Hand PoseICCV 2023
AffordPoseNet has low affordance accuracy for pull (0%) and twist (53.13%) affordances because the target parts (e.g., bag zippers, bottle lids) are small and generated hands often contact nearby wrong regions
Nearest existing work
- AffordPose: A Large-Scale Dataset of Hand-Object Interactions with Affordance-Driven Hand PoseICCV 2023
- AffordGrasp: Cross-Modal Diffusion for Affordance-Aware Grasp SynthesisCVPR 2026
- Understanding Everyday Hands in Action From RGB-D ImagesICCV 2015
- AdaAfford: Learning to Adapt Manipulation Affordance for 3D Articulated Objects via Few-Shot InteractionsECCV 2022
- TOUCH: Text-guided Controllable Generation of Free-Form Hand-Object InteractionsICLR 2026
- Beyond the Contact: Discovering Comprehensive Affordance for 3D Objects from Pre-trained 2D Diffusion ModelsECCV 2024
- AffordMatcher: Affordance Learning in 3D Scenes from Visual SignifiersCVPR 2026
- Learning Precise Affordances from Egocentric Videos for Robotic ManipulationICCV 2025
- Leverage Interactive Affinity for Affordance LearningCVPR 2023
- Task-Oriented Human Grasp Synthesis via Context- and Task-Aware DiffusersICCV 2025
- Grounding 3D Object Affordance from 2D Interactions in ImagesICCV 2023
- GanHand: Predicting Human Grasp Affordances in Multi-Object ScenesCVPR 2020
- Hand-Object Contact Consistency Reasoning for Human Grasps GenerationICCV 2021
- Where2Act: From Pixels to Actions for Articulated 3D ObjectsICCV 2021
- Human Hands As Probes for Interactive Object UnderstandingCVPR 2022
Related open problems
Hand-Object Interaction & Affordance
Robust Hand-Object Interaction Learning Beyond Cascaded Parametric Pose Estimators
Downstream hand-object interaction (HOI) and affordance models are universally bottlenecked by a strict reliance on upstream off-the-shelf 3D hand pose estimators (e.g., FrankMocap, HaMeR, MANO parameterizations) and bounding-box detectors. When these upstream tools encounter severe object occlusions, fast manipulation dynamics, or egocentric perspective distortion, their estimation errors cascade directly into downstream HOI training, causing catastrophic mis-projections and synthetic contact artifacts. Because downstream methods require clean parametric 3D meshes as preconditions, massive repositories of in-the-wild manipulation videos cannot be utilized without fragile filtering heuristics that discard over half the data.
Hand-Object Interaction & Affordance
Robust 3D Hand-Object Pose and Interaction Estimation Under Upstream 2D Segmentation and Localization Noise
Current hand-object interaction and 3D hand pose estimation models rely heavily on upstream 2D bounding box crops, hand segmentation masks, or pre-extracted interaction regions. In real-world egocentric scenes, upstream 2D segmenters and detectors degrade severely due to object occlusions, motion blur, and cluttered backgrounds, causing downstream 3D estimation to fail or hallucinate. Because models are trained and benchmarked assuming near-ideal 2D crops or masks, existing pipelines cannot operate reliably in autonomous, in-the-wild video streams.
Hand-Object Interaction & Affordance
Benchmarking Hand-Object Interaction Models Across Diverse Hand Morphologies and Embodiments
Existing hand representation and hand-object interaction models are predominantly evaluated on small demographic cohorts or fixed kinematic setups (e.g., standard anthropomorphic models like MANO or fixed robotic hands like ShadowHand). Because existing pipelines implicitly bind kinematics, palm-to-finger ratios, or subject identities into their representations, their performance on unseen human hand proportions and non-standard robotic end-effectors remains untested. As a result, it is unknown whether current interaction and affordance estimators degrade gracefully or fail catastrophically when applied outside their narrow training morphologies.
Hand-Object Interaction & Affordance
Hand-Object Affordance Estimation Beyond Clean Pre-Scanned Scenes and Predefined 3D Grids
Current methods for 3D hand-object interaction and affordance estimation depend strictly on clean geometric preconditions: high-fidelity 3D scans, pre-detected object instances, static overhead views, or tightly predefined 3D bounding grids. In practical manipulation and augmented reality scenarios, environments are unsegmented, viewpoints are dynamic, and point clouds are noisy and incomplete. Relying on manually tuned grid volumes and pre-extracted meshes prevents existing models from operating directly on raw, uncurated sensor streams.