CMDA: Cross-Modal and Domain Adversarial Adaptation for LiDAR-Based 3D Object Detection

Keyphrases

• Cross-modal 100%
• LiDAR 100%
• Cross-domain 100%
• 3D Object Detection 100%
• Domain Adversarial Adaptation 100%
• Object Detection Algorithm 75%
• Target Domain 50%
• Unsupervised Domain Adaptation 50%
• Domain Gap 50%
• Camera Image 25%
• Performance Gain 25%
• Adaptive Features 25%
• Unseen 25%
• Self-training 25%
• Large-scale Benchmark 25%
• Learning Strategies 25%
• Data Distribution 25%
• Training Base 25%
• Image Modality 25%
• Source Data 25%
• Art Performance 25%
• Visual Semantics 25%
• Domain-invariant Feature 25%
• Domain Adaptation Method 25%
• Domain Adaptive 25%
• KITTI 25%
• Training Data Distribution 25%
• Semantic Cues 25%
• NuScenes 25%
• Bird's Eye View 25%
• Waymo 25%

Computer Science

• 3D Object Detection 100%
• Data Distribution 50%
• Unsupervised Domain Adaptation 50%
• Detection Method 25%
• Training Data 25%
• Performance Gain 25%
• Art Performance 25%
• View Representation 25%
• Invariant Domain 25%