Fire! Fire! Fire! Who hasn’t yet owned the “Orange Pi” that is currently all the rage among developers? What happens when the brand new “Orange Pi” meets versatile developers?

Recently, the first phase of the “Orange Pi Free Experience Program” jointly launched by the ITPUB community, Orange Pi, and Ascend AI has successfully concluded. The community gathered a large number of developers’ attention through innovative methods such as “Orange” inviting talents → “Orange” winning the chase → thinking of “Orange” and provided an open communication learning platform and practical opportunities for developers.

During this event, developers showcased their creativity and unique skills, deeply experiencing the powerful performance and rich features of the Orange Pi Alpro development board, and presented their great ideas and applications on the Orange Pi Alpro development board, ultimately sharing their application practices and insights. Let me take you to have a look~

Orange Pi Alpro, Starting the Journey of Artificial Intelligence

Our developers are truly impressive! As of the end of the event, hundreds of developers participated, with nearly a hundred Orange Pi Alpro development boards ordered. We collected over 44 experience practice videos and more than 13 experience report articles from developers participating in the “Playing with Orange Pi” program, with nearly a thousand community discussions.

It must be said that the Orange Pi Alpro development board, with its unique Ascend AI technology route and rich interface design, has attracted the attention of many developers. Its provided 8/20 TOPS powerful computing power allows developers to easily handle various complex AI application scenarios.

In terms of application practice, developers have made various innovative attempts based on the Orange Pi Alpro development board:

Some developers utilized the development board for image classification and recognition tasks, achieving high-precision image recognition results through training and optimizing models:

For example, the Ascend Yolov8n face detection quick migration case: the developer used the YOLOv8n model, first cloning the code to the Orange Pi AIPro development board; next, downloading yolov8n-face.pt; the third step is to convert yolov8n-face.pt to yolov8n-face.onnx; the fourth step is to convert yolov8n-face.onnx to yolov8n-face.om; the final step is to perform model inference.

After executing the above inference commands, the system processes the image and outputs results, usually drawing a bounding box around detected faces, indicating the locations of faces successfully detected by the model. This is crucial for verifying the performance and accuracy of the model on specific hardware (such as the Ascend 310 on the Orange Pi AIPro development board). You should be able to see clearly marked face areas, marking the successful completion of the entire migration and deployment process.

For instance, a developer created a simple camera application that automatically generates a white background using the Orange Pi AIpro development board, which I believe is a practical and efficient tool that provides great convenience for photography enthusiasts and image processing workers.

The powerful computing capability and efficient image processing speed of the Orange Pi AIpro development board make this camera application very smooth in actual use. This technology not only eliminates cumbersome manual cutout and post-processing steps but also ensures image quality, allowing each photo to present a good visual effect, closely aligned with real life.

Furthermore, on the Orange Pi AIpro, developers can perform image reading, horizontal flipping, and grayscale transformation. By writing the corresponding code, these functions can be easily achieved.

When executing these operations on the Orange Pi AIpro, developers fully utilized its powerful computing capabilities and optimized image processing libraries. These functions have broad application value in image processing, computer vision, and machine learning fields. By mastering these technologies, we can better utilize the Orange Pi AIpro platform for image processing-related work and achieve good results in practical applications.

Some developers applied the development board in the field of voice recognition, achieving efficient speech-to-text functionality through deep learning algorithms:

For example, a smart security alarm system case sharing: there are elderly people at home, and surveillance is installed. Generally, one would only check the surveillance when they remember, which often leads to missing many moments. Therefore, the goal is to use algorithms to replace manual recording of moments when someone is at home.

The developer used the Orange Pi AIpro development board for a vision-based intelligent monitoring system project. By reading the images from existing network cameras at home, the developer added polygonal area restrictions to limit the detection range. Human shape recognition was performed on the images within the detection range.

The developer stated that since the Orange Pi AIpro is an AI development board, it is well-suited for solving such problems with AI algorithms through the integrated Ascend 310 SOC. The algorithm chosen for detecting human shapes was yolov5, and the AIpro has perfect compatibility with yolov5, fully utilizing the NPU’s 8T computing power. After recognizing human shapes, it calculates alarm delays and triggers alarms. A web service is built on a cloud server or locally on the AIpro, and when the recognition end detects a human shape, it sends alarm information to the web backend via POST, which receives the alarm information for database recording and alarm display.

Additionally, on the Orange Pi AIpro, voice recognition can be performed: the developer stated that performing voice recognition on the Orange Pi AIpro is a highly practical function. During the operation, we started the voice recognition function through specific directories and scripts, achieving accurate recognition and conversion of voice signals.

The implementation of the voice recognition function not only showcases the powerful capabilities of the Orange Pi AIpro in the field of artificial intelligence but also reflects its broad application prospects in various fields such as the Internet of Things, smart homes, and robot control. Through voice recognition, we can interact with devices more conveniently, enhancing user experience and work efficiency.

Some developers utilized the development board for natural language processing tasks, such as text classification and sentiment analysis, achieving satisfactory results. For example: local large models for dialogue robots; scene text detection (outdoor scenes), etc.

Moreover, the Orange Pi Alpro development board can be widely applied in AI teaching and training, AI algorithm verification, smart cars, robotic arms, edge computing, drones, artificial intelligence, cloud computing, AR/VR, smart security, smart homes, smart transportation, and more.

Among these, the intelligence and creativity of our developers are also evident, such as in the medical field, where developers shared applications for detecting mild and severe fractures in children’s wrist CT images (including foreign object intrusion), brain hemorrhage detection, lymphocyte extraction and counting, etc.

In the transportation field, developers shared parking space detection, pedestrian detection in automatic driving dashcams, and more.

I believe that using the Orange Pi AIpro for parking space detection is very practical and fully utilizes the powerful computing capabilities of the Orange Pi AIpro development board and the advantages of Ascend AI technology, enabling real-time and accurate identification of available parking spaces. Through deep learning algorithms and image recognition technology, developers can train efficient parking space detection models to quickly judge and update the occupancy status of parking spaces. This not only improves the management efficiency and service level of parking lots but also provides car owners with a more convenient and comfortable parking experience.

In the manufacturing field, developers shared industrial scene spray font recognition, based on an AI industrial scene spray font detection system that can work with ABB robotic arms to achieve automatic recognition and detection of spray fonts.

In addition to traditional industry fields, our developers have also produced numerous innovative applications, such as using AI for real-time video background replacement, detecting dozens of wild animals in wildlife protection areas, automating equipment detection for wheat maturity by local agricultural bureaus, and collecting data on rare wild plant diversity, etc.

It is not difficult to see that whether in AI edge computing, deep visual learning, video stream AI analysis, or natural language processing, the Orange Pi Alpro development board can demonstrate outstanding performance.

During the event, developers fully experienced the convenience of developing AI applications through the Ascend CANN software stack’s AI programming interface. AscendCL, as the Ascend computing open programming framework, provides developers with rich APIs and powerful functional support. Developers utilized AscendCL for deep learning inference calculations, graphics and image preprocessing, which not only improved development efficiency but also significantly enhanced application performance.

In addition to application practices, developers also expressed their recognition and appreciation for the Orange Pi Alpro development board. They believe that the Orange Pi Alpro development board is not only powerful but also has excellent expandability, capable of meeting the needs of various complex AI application scenarios. At the same time, the ease of use and powerful features of the Ascend CANN software stack and AscendCL programming framework have also greatly benefited developers.

Orange Pi AIpro, Born for AI

Why is the Orange Pi AIpro development board so favored by developers? It is known as the industry’s first AI development board based on deep research by Ascend, the Orange Pi AIpro excels in appearance, performance, and technical service support.

OrangePi AIpro (8T) adopts the Ascend AI technology route, specifically a 4-core 64-bit processor + AI processor, integrated graphics processor, supporting 8TOPS AI computing power, with 8GB/16GB LPDDR4X, and can connect external 32GB/64GB/128GB/256GB eMMC modules, supporting dual 4K HD output.

The Orange Pi AIpro features a rich array of interfaces, including two HDMI outputs, GPIO interfaces, Type-C power interfaces, an M.2 slot supporting SATA/NVMe SSD 2280, a TF slot, a gigabit network port, two USB3.0, one USB Type-C 3.0, one Micro USB (for serial printing and debugging), two MIPI cameras, one MIPI screen, and a reserved battery interface, making it widely applicable in fields such as artificial intelligence, covering various industries in AIoT.

In terms of operating systems, the OrangePi AIpro can run Ubuntu and openEuler operating systems, meeting the needs of most AI algorithm prototype verification and inference application development.

Conclusion

Looking back at the “Orange Pi Free Experience Program” event, the community not only provided developers with a learning and practice opportunity but also promoted the popularization and application of AI technology. In the future, we look forward to seeing more excellent applications based on the Orange Pi Alpro development board emerge, injecting new vitality into the development of the AI field. At the same time, we, the ITPUB community, Orange Pi, and Ascend AI, hope to continue hosting more similar events to create more learning and communication opportunities for developers.

For more event content, please click: https://zt.itpub.net/topic/OrangePi/

For more Orange Pi content, please click: http://www.orangepi.cn/

• OpenAI flagship model GPT-4o has arrived, enabling seamless communication with humans, completely free!

• The year of microservices “retrograde”, Google has reduced microservices costs by 9 times!

• 600,000 people’s $125 billion cannot be withdrawn, Google also made a mistake!

• What is it like to trigger an operation and maintenance accident yourself?

• Are you running MySQL in Docker? Congratulations, you are out of a job!

• Due to accidental database deletion, this software industry’s Hermes encountered the biggest failure in its 5-year history.

• Unlocking the secrets of modern data architecture, the “Real-time Data Warehouse Selection Guide” e-book (with download) is officially online!

• DTCC2023: Database industry trends, cutting-edge technologies, and market operation strategies interpretation (with conference PPT download)