RA-MOSAIC: Resource Adaptive Edge AI Optimization over Spatially Multiplexed Video Streams

Computer Science

• High Throughput 100%
• Computational Resource 100%
• Computational Load 100%
• Detection Accuracy 100%
• Criticality 100%
• Object Detection 100%
• Deep Neural Network 100%
• Video Transmission 100%
• Observed Object 100%
• Resource Environment 100%
• Multiple Camera 100%
• license plate recognition 100%
• Benchmark Task 100%
• Packing Strategy 100%
• Edge AI 100%
• Artificial Intelligence 100%
• Graphics Processing Unit 100%
• Drone 100%

Keyphrases

• Multiplexing 100%
• Video Stream 100%
• Edge AI 100%
• Resource-adaptive 100%
• Adaptive Edge 100%
• Region of Interest 50%
• Edge Devices 50%
• Bandwidth Adaptive 50%
• Throughput Gain 25%
• Camera Sensor 25%
• Vision Perception 25%
• High-throughput 12%
• Accuracy Improvement 12%
• Detection Accuracy 12%
• Computational Load 12%
• Dynamic Change 12%
• Selected Region 12%
• Downsampling 12%
• Uniform Grid 12%
• Computation Overhead 12%
• Demand-side Resources 12%
• Batch Size 12%
• AI-based Methods 12%
• Object Detection 12%
• Bandwidth-constrained 12%
• Bin Packing 12%
• Multi-tenancy 12%
• Load Adaptive 12%
• Task Success 12%
• Wireless Environment 12%
• First Come First Served 12%
• Deep Neural Network 12%
• Edge Nodes 12%
• Multiple Cameras 12%
• Drone 12%
• Transmission Method 12%
• Task Processing 12%
• Pedestrian Detection 12%
• Mixed Resolution 12%
• Environment Resources 12%
• NVIDIA Jetson TX2 12%
• Limited Computational Resources 12%
• Benchmark Tasks 12%
• Packing Strategy 12%
• Automatic License Plate Recognition 12%
• Adaptive Bins 12%