Banana Pi BPI-F4: Powered by Sunplus SP7350, 4.6TOPS NPU  AI Solution for Robotics and Smart Devices

• Quick Start: https://sunplus.atlassian.net/wiki/spaces/C3/pages/2212036657/User+Manual+of+SP7350+Mini+Control+Board+MCB
• SDK: https://sunplus.atlassian.net/wiki/spaces/C3/overview
• Tips

  • Most Clicked Link

    Hardware Kit: User Manual of C3V IO Board

    Software BSP: Downloading and Compiling Code

    AI Engine: NN User Guide Overview

    NPU Benchmark: NN Models' Performance with NPU

Introduction

Banana Pi BPI-F4 industrial control board, powered by the Sunplus SP7350 chip, delivers a remarkable 4.6TOPS AI power, tailored for AI vision applications. It features a quad-core 1.8GHz ARM Cortex-A55 CPU and an independent ARM Cortex-M4 MCU, along with abundant peripheral interfaces such as a MIPI-RX camera interface for high-speed data transfer. Perfect for robots, drones, smart cameras, video conferencing systems, driver monitoring, and more, it drives edge AI and AIoT application development. Using advanced 12nm technology, it offers low power consumption and compact size, making it ideal for building small-form-factor and energy-efficient devices.

Features

• [High-Performance AI Chip] The Sunplus SP7350 chip, boasting 4.6TOPS AI computing power, is specifically designed for AI vision applications, providing robust performance for edge AI and AIoT tasks.
• [Powerful CPU and MCU Combination] Equipped with a quad-core 1.8GHz ARM Cortex-A55 CPU and an independent ARM Cortex-M4 MCU, it efficiently handles computer vision, video, image, and audio processing.
• [Rich Peripheral Interfaces] It integrates a MIPI-RX camera interface with automatic pin detection, signal switching, and error detection, ensuring smooth and user-friendly connectivity for high-speed data transfer.
• [Advanced Process and Low Power Consumption] Fabricated using advanced 12nm chip technology, it offers a small form factor and low power consumption, ideal for building compact and energy-efficient devices.
• [Wide Range of Applications] It is suitable for various fields such as robotics, drones, AMR/AGV, surveillance, inspection, counting, people tracking, and human-machine interfaces for gaming or touchless control.
• [Compatible with Multiple AI Frameworks] Supports popular AI frameworks like Caffe, TensorFlow, PyTorch, TFLite, DarkNet, and ONNX, as well as over 50 AI models such as YOLO - v5/V8/V10, meeting diverse development needs.
• [Versatile Development Platform] Provides an open Linux platform, making it convenient for developers to quickly deploy AI applications and accelerate product development cycles.

Shipping List

Reference to the picture.

Applications

Specification

Detailed Descriptions