End-to-End FPGA-based Object Detection Using Pipelined CNN and Non-Maximum Suppression

Keyphrases

• Autonomous Driving 14%
• Bounding Box 28%
• Computer Vision Tasks 14%
• Confidence Score 14%
• Convolutional Layer 42%
• Convolutional Neural Network 100%
• Convolutional Neural Network Accelerator 14%
• Custom Hardware 14%
• End System 14%
• Extracting Features 14%
• Feature Extraction 14%
• FPGA-based 100%
• Frames per Second 14%
• Fully pipelined 14%
• Hardware Implementation 28%
• High-score 14%
• High-throughput 14%
• Intel 14%
• Latency 14%
• Latency Overhead 28%
• Low Latency 28%
• Maximum Operating Frequency 14%
• Minimal Latency 14%
• MobileNetV1 14%
• Non-maximum Suppression 100%
• Object Detection 100%
• Object Detection System 28%
• Partially Overlapped 14%
• Robotics 14%
• Single Shot Detector 42%
• Smart Surveillance 14%
• Stratix 10 14%
• Throughput Degradation 14%
• Two-component 14%

Computer Science

• Autonomous Driving 20%
• Computer Vision Task 20%
• Confidence Score 20%
• Convolution Layer 40%
• Convolutional Layer 20%
• Convolutional Neural Network 100%
• Custom Hardware 20%
• Feature Extraction 20%
• Hardware Implementation 40%
• High Throughput 20%
• Object Detection 100%
• Operating Frequency 20%
• Smart Surveillance 20%