Shypd.aiThe largest open index of AI toolsNeural-Network-Accelerator is an open-source AI tool that implements a Convolutional Neural Network accelerator on FPGA. It enables the inference of YOLOv8 in 60ms on Xilinx devices.
Trending
About
Neural-Network-Accelerator is an open-source project focused on implementing a Convolutional Neural Network (CNN) accelerator on FPGA, specifically targeting Xilinx devices (xczu7ev-ffvc1156-2-i). The project uses YOLOv8 as its primary implementation object, achieving an inference time of 60ms for YOLOv8n. The accelerator supports various operator operations crucial for CNNs, including convolution, residual sum, upsampling, and pooling. Its design emphasizes efficient reusability, allowing other neural networks composed of these operators to be compiled onto the accelerator. The tool provides simulation results and instructions for both simulation and hardware implementation, making it valuable for developers working on embedded AI applications and hardware acceleration.
Capabilities
Pricing & Plans
Open Source
Free
FAQs
Trending
