Research map / Vision-language and multimodal / Cross-modal learning
ModDrop: Adaptive Multi-Modal Gesture Recognition
Natalia Neverova, Christian Wolf, Graham Taylor, Florian Nebout
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 ModDrop to dropout training (ablation) | tree-structured MNIST net · MNIST (1 segment covered) | test error %: 10.74 -> 2.30 | Improved |
| add ModDrop to dropout training (ablation) | tree-structured MNIST net · MNIST (3 segments covered) | test error %: 68.03 -> 24.88 | Improved |
| add ModDrop (ablation) | MNIST net with all segments visible · MNIST | test error %: 1.02 -> 1.02 | No clear effect |
| add ModDrop (ablation) | MNIST net under 50% pepper noise on all segments · MNIST | test error %: 7.27 -> 6.42 | Improved |
| add modality-wise pretraining (ablation) | MNIST multi-modal net · MNIST (4 quarters as modalities) | test errors: 142 -> 118 | Improved |
| add dropout on inputs after pretraining (ablation) | MNIST multi-modal net · MNIST (4 quarters as modalities) | test errors: 118 -> 102 | Improved |
| add ModDrop on top of pretraining and dropout (ablation) | MNIST multi-modal net · MNIST (4 quarters as modalities) | test errors: 102 -> 102 | No clear effect |
| add ModDrop to pretraining, dropout and initialization (ablation) | ModDrop network · ChaLearn 2014 LAP + audio (validation) | accuracy %: 96.77 -> 96.81 | Improved |
| add shared-layer initialization with pretrained blocks (ablation) | ModDrop network · ChaLearn 2014 LAP + audio (validation) | accuracy %: 96.31 -> 96.77 | Improved |
| add modality pretraining (ablation) | network without dropout · ChaLearn 2014 LAP + audio (validation) | accuracy %: 91.94 -> 94.96 | Improved |
| add ModDrop when the mocap channel is missing (ablation) | dropout network · ChaLearn 2014 LAP + audio | accuracy %: 38.41 -> 92.82 | Improved |
| add gesture localization module (ablation) | ModDrop improved · ChaLearn 2014 LAP visual | Jaccard index: 0.821 -> 0.868 | Improved |
Closest papers by meaning
- Online Detection and Classification of Dynamic Hand Gestures With Recurrent 3D Convolutional Neural Network (CVPR 2016 · similarity 0.87)
- Improving the Performance of Unimodal Dynamic Hand-Gesture Recognition With Multimodal Training (CVPR 2019 · similarity 0.87)
- Challenges in multimodal gesture recognition (JMLR 2016 · similarity 0.87)
- Learning Adaptive Hidden Layers for Mobile Gesture Recognition (AAAI 2018 · similarity 0.86)
- X-Fi: A Modality-Invariant Foundation Model for Multimodal Human Sensing (ICLR 2025 · similarity 0.85)
- SocialGesture: Delving into Multi-person Gesture Understanding (CVPR 2025 · similarity 0.85)
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.
