HiMODE: A Hybrid Monocular Omnidirectional Depth Estimation Model

Engineering

• Extracted Feature 100%
• Performance Degradation 100%
• Input Image 100%
• Conducted Experiment 100%
• High Resolution 100%
• Limitations 100%
• Computational Cost 100%
• Supplementary Material 100%
• Convolutional Neural Network 100%

Keyphrases

• Omnidirectional Depth Estimation 100%
• Transformer Encoder 100%
• Estimation Model 100%
• Depth Map 50%
• Spatial-temporal 25%
• Computational Cost 25%
• Visual Quality 25%
• Monocular Depth Estimation 25%
• Ablation Study 25%
• CNN-Transformer 25%
• Small Objects 25%
• Ground Truth Depth Map 25%
• Distortion Cost 25%
• Attention Layer 25%
• High-resolution Feature 25%
• Number of Parameters 25%
• Art Performance 25%
• Feature Pyramid Network 25%
• Feature Extracting 25%
• Self-attention 25%
• Cross-attention 25%
• Encoder-decoder Architecture 25%
• Residual Block 25%
• Spatial Residuals 25%
• Performance Degradation 25%
• Network Applications 25%

Computer Science

• Depth Estimation 100%
• Convolutional Neural Network 25%
• Deep Feature 25%
• Visual Quality 25%
• Art Performance 25%
• Attention (Machine Learning) 25%
• Computational Cost 25%
• Performance Degradation 25%
• Supplementary Material 25%