[The 3rd Xuantie Cup RISC-V Application Innovation Competition] LicheePi 4A+ portable mobile endoscope
[Copy link]
[The 3rd Xuantie Cup RISC-V Application Innovation Competition] LicheePi 4A+ portable mobile endoscope
Author: Azuma Shimoe
Project Background ( 100-200 words )
Both medical and industrial fields are moving towards specialization. Medical electronics is becoming more and more popular. Endoscopes play an important role in medical diagnosis and large industrial equipment inspection. However, medical and industrial equipment are often large and cannot be moved. If we can make an endoscope kit that is both compact and portable with a mobile power supply and has intelligent image recognition function, it will fill a gap in low price and low cost.
Introduction ( 100-200 words)
This design uses the RISC - V development board to create a mobile endoscope, and uses open source code to train a device that can recognize image objects. The training data can basically recognize the image. At present, the hardware has been built and the lens capture can be started smoothly. The algorithm framework for target recognition has also been built, and the image recognition training is still under development (to be supplemented later). I have to say here that I didn’t find a good python IDE for RISCV , and I also missed some software, which led to twists and turns in my development process. I still need a lot of experts and manufacturers to develop and transplant it.
System block diagram (combined with pictures and text)
Main components:
Display screen * 1
Development board * 1
Endoscope * 1
Mobile power supply * 1
Functional description of each part (combined with pictures and text)
The development board is mainly used to control the entire system, and the display screen can display the endoscope image in real time; the hardware structure is relatively simple. However, the box window opening has become the most labor-intensive. It would be great if the official could provide a box with various window openings. I would consider buying one as my own NAS.
Family portrait
Video presentation (video description + link)
111_智能字幕
The video shows how to assemble the kit, a simple trial, and actual shots of the effects in a small space. Overall, it is good and clear.
The picture shows the cables inside the narrow pipe
Project source code
usb_test.py
(296 Bytes, downloads: 4)
Published blog post (attached with title and Pingtouge release link)
[Xuantie Cup 3rd RISC-V Application Innovation Competition] LicheePi 4A+ model construction
https://www.xrvm.cn/community/post/detail?id=424067148626 5856000
[Xuantie Cup 3rd RISC-V Application Innovation Competition] LicheePi 4A+ portable mobile endoscope hardware construction completed
https://www.xrvm.cn/community/post/detail?id=4244179381871775744
Project summary ( summarize the work completion, technical exploration, innovation points, etc.)
Overall, the fun lies in its portability, plug-and-play, and suitability for a variety of scenarios. Of course, the USB-endoscope itself can be plug-and-play with a mobile phone, and the board plus the display is several times larger than the mobile phone. However, few mobile apps have recognition functions, so using a computing board with a built-in GPU to create an intelligent recognition endoscope is very useful in special medical and industrial scenarios.
The author tried to put the ear into the ear to check the inside (please do not try it privately )
The author of the post personally tried it and gently reached into the ear to see the dirt inside. It felt similar to cleaning the ears outside. Just look inside and gently wipe it with an ear spoon and cotton swab and it will basically be done.
Minors are prohibited from attempting dangerous actions.
To be continued, save for later, mainly equipped with intelligent recognition algorithms
I have encountered a problem with the official YOLOX model and I am still working on it. I am not knowledgeable enough and have not figured it out yet. I hope some experts can give me some advice.
An error message is reported when the model is started. It has not been resolved yet.
IX . Others
|