Survey maps dynamic neural networks for computer vision and sensor fusion

This survey paper provides a comprehensive overview of Dynamic Neural Networks (DNNs), focusing on their application in computer vision and multi-modal sensor fusion. It addresses the challenge of deploying large models on embedded devices by explaining how DNNs adapt computation levels based on input complexity, unlike static optimization methods. The paper categorizes DNNs by their adaptive components—output, computation graph, or input—and highlights their potential benefits for sensor fusion tasks, such as improved adaptivity and noise reduction. AI

Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →

IMPACT Provides a unified taxonomy and repository for dynamic neural network research, potentially accelerating development in adaptive AI systems for resource-constrained environments.

RANK_REASON This is a survey paper published on arXiv detailing research in Dynamic Neural Networks.

Read on arXiv cs.CV →

• Dynamic Neural Networks
• Computer Vision
• Multi-modal Sensor Fusion
• model compression
• embedded devices
• pruning
• quantization
• sensor fusion
• arXiv

• paper
• infra