Attention-Aware Feature Aggregation for Real-Time Stereo Matching on Edge Devices

Keyphrases

• Depth Estimation 100%
• Edge Devices 100%
• Disparity Map 100%
• Real-time Stereo Matching 100%
• Feature Aggregation 100%
• Attention-aware 100%
• 3D Convolutional Neural Network (3D CNN) 66%
• CNN Architecture 66%
• Multi-scale Feature Maps 66%
• Stereo Image Pair 33%
• Feature Extraction 33%
• Convolutional Neural Network 33%
• Cost Volume 33%
• Stereo Pair 33%
• Time Application 33%
• Feature Map 33%
• Computational Power 33%
• Computational Resources 33%
• Informative Features 33%
• Disparity Estimation 33%
• Efficient Extraction 33%
• NVIDIA Jetson TX2 33%
• Aggregated Features 33%
• Feature Aggregation Module 33%
• Representational Capacity 33%
• Real-time Inference 33%

Computer Science

• Stereo Matching 100%
• Depth Estimation 100%
• Feature Map 100%
• 3-Dimensional Convolutional Neural Networks 66%
• Experimental Result 33%
• Convolutional Neural Network 33%
• stereo-image 33%
• stereo pair 33%
• Real-Time Application 33%
• Computational Power 33%
• Computational Resource 33%
• Representational Capacity 33%
• Feature Extraction 33%