The 'Internet+' College Students' Innovation and Entrepreneurship Competition

• Hardware Platform

Shaolin Pi Development Board

Shaolin Pi development board is a high-performance platform based on BM1684, providing around 20 TOPS of computing power. With BM1684 as its core component, the board features a fully autonomous and controllable processor, delivering exceptional computing power and multi-channel video encoding and decoding capabilities. It supports 3 mini-PCIe and 4 USB interfaces, allowing for the expansion of various peripheral modules. It optimizes configurations according to scene requirements, balancing cost-effectiveness, energy efficiency, and functionality. The hardware ecosystem is extensive, accommodating diverse peripheral connections. It boasts support for a rich software development ecosystem, compatible with mainstream deep learning frameworks. The 'Shaolin Pi' core board can expand connectivity to screens, keyboards, mice, cameras, headphones, VR, and other devices. It enables the creation of an all-encompassing edge computing workstation for various AI experiments. Additionally, it can be embedded into unmanned vehicles and drones, enabling edge computing for mobile terminals. Leveraging the Shaolin Pi development board, our company has developed the KT001 smart car and S550 deep learning drone. Participants can also request the deep learning car as a hardware platform to complete their own parameterized works for the competition.

The expansion modules supported by the Shaolin Pi development board are as follows:

Shaolin Pi Development Materials：https://www.sophgo.com/curriculum/description.html?category_id=6

• Milk-V Duo

Milk-V Duo is an ultra-compact embedded development platform based on the CV1800B processor (RISC-V architecture, C906@1GHz + C906@700MHz). It supports 64MB RAM and can expand to achieve 10/100Mbps Ethernet. This board can run Linux and RTOS systems, providing a reliable, low-cost, high-performance platform for professionals, industrial ODMs, AIoT enthusiasts, DIY enthusiasts, and creators.

Development materials for the Milk-V Duo:

1. Docker：https://hub.docker.com/repository/docker/dreamcmi/cv1800-docker/general

2. GitHub：https://github.com/milk-v/duo-manifest

3. SDK：https://developer.sophgo.com/thread/471.html

4. Docs：https://milkv.io/docs/duo

• Milk-V Pioneer

Milk-V Pioneer is a development board based on the SOPHON SG2042 processor, designed in a standard mATX form factor. Leveraging PC-like interfaces and PC industrial compatibility, it provides a native RISC-V development environment and a RISC-V desktop experience. It's the preferred choice for RISC-V developers and hardware pioneers to explore cutting-edge RISC-V technology.

Milk-V Pioneer Board

Milk-V Pioneer Development Materials: http://milkv.io/docs/pioneer/getting-started

Participants can select the expansion modules mentioned above to complete their competition projects or design their own expansion modules.

• Software Platform

SophonSDK is a deep learning SDK customized by SOPHGO based on its independently developed AI processor. It encompasses the necessary capabilities during the neural network inference phase, such as model optimization and efficient runtime support, providing an easy-to-use, efficient end-to-end solution for deep learning application development and deployment.

SophonSDK consists of a Compiler and Library:

The Compiler is responsible for offline compilation and optimization of neural network models trained under third-party deep learning frameworks, generating the required BModel for the final runtime. It currently supports Caffe, Darknet, MXNet, ONNX, PyTorch, PaddlePaddle, TensorFlow, and other frameworks.

The Library includes BM-OpenCV, BM-FFmpeg, BMCV, TPURuntime, BMLib, and other libraries, which drive hardware components like VPP, VPU, JPU, TPU, completing video/image encoding/decoding, image processing, tensor operations, model inference, and other functions for users in deep learning application development.

SOPHGO provides SDK-related materials for participants to study:

1. Documentation Center: https://developer.sophgo.com/site/index/material/30/all.html
2. Video Tutorials: https://developer.sophgo.com/site/index/course/all/all.html
3. Development Guide: https://sophgo-doc.gitbook.io/sophonsdk3

• Technical Specifications

a) Utilize open-source or self-developed technologies and agree to open-source the design after the competition.

b) Constructing a prototype solution is a fundamental requirement.

• Submission Standards

a) Prototype solution (including circuit schematics source files and complete source code)

b) Development documentation (recording the development process)

c) Demonstration video (explaining the solution)

• Task List

a) Topics 1-3 utilize the SOPHON Shaolin Pi development board, topics 4-5 use the Milk-V Duo development board, and topic 6 uses the Milk-V Pioneer complete machine or another development board equipped with SG2042 to build the prototype solution.

b) Debugging and development

c) Verify that the solution meets functional requirements

d) Optimize the design and document the entire development process as a development document

Building a prototype solution is the basis for further expansion. Additional scoring criteria are as follows:

• Topic 1:

a) Innovation: Extra points for innovative design and implementation in areas like face detection and recognition algorithms, access control strategies, data storage, and management.

b) Security Assurance: Implementing multiple identity authentication and anti-spoofing measures in the facial recognition access control system for enhanced security can earn additional points.

c) User Experience: Well-designed user interfaces providing simple operations and a friendly user experience will receive extra points.

• Topic 2:

a) Accuracy and Efficiency: Systems for depth estimation require high precision in generating depth images and point clouds while maintaining good computational efficiency and speed. Exceptional performance in this area earns additional points.

b) Model Training and Optimization: Thorough model training and optimization to enhance the accuracy and efficiency of the depth estimation system. Employing advanced deep learning techniques like adaptive learning, transfer learning, with good results will get extra points.

c) Application Scenario Expansion: Designing and optimizing the deep estimation technology for specific scenarios like autonomous driving, robot navigation, virtual reality, etc., and achieving good results will earn additional points.

• Topic 3:

Object Detection and Tracking: Detecting and tracking objects in panoramic images or videos and outputting detection results and tracking trajectory information will receive additional points.

• Topic 4:

a) Product-oriented appearance and structural design.

b) Extra and innovative functionalities apart from the basic ones implemented.

c) Innovative interaction.

d) Detailed development documentation.

• Topic 5:

a) Stable flight control.

b) Use of high-illumination lighting equipment.

c) Implementing remote real-time control.

d) Excellent appearance and structural design.

• Topic 6:

a) Product-oriented appearance and structural design.

b) Implementing additional and innovative functionalities alongside the basic ones.

c) Innovative interaction.

d) Detailed development documentation.