A High-Throughput Full-Dataflow MobileNetv2 Accelerator on Edge FPGA

Keyphrases

• High-throughput 100%
• MobileNetV2 100%
• Dataflow Architecture 60%
• Feature Map 40%
• Integer Quantization 40%
• Resource Allocation Strategy 40%
• Latency 20%
• Parallelization 20%
• 8-bit 20%
• Accuracy Loss 20%
• FPGA Acceleration 20%
• Residual Connection 20%
• Memory Resource 20%
• ZCU102 20%
• Lightweight Neural Network 20%
• Intermediate Features 20%
• Computing Resource Allocation 20%
• Xilinx 20%
• Quantization Accuracy 20%
• Edge Computing Applications 20%
• On-chip Memory 20%

Computer Science

• High Throughput 100%
• MobileNetV2 100%
• Quantization (Signal Processing) 60%
• Resource Allocation 40%
• Allocation Strategy 40%
• Experimental Result 20%
• Neural Network 20%
• Edge Computing 20%
• Parallelism 20%
• Computing Resource 20%
• Memory Resource 20%