OpenMV Review – Machine Vision Camera Module


Dear friends welcome to another video tutorial! In this video, we are going to take a first
look at the OpenMV camera module, a board that wants to become the Arduino of Machine
Vision. Let’s get started! Hello, guys, I am Nick and welcome to educ8s.tv
a channel that is all about DIY electronics projects. In this channel, I share everything about
the projects I build to help you develop similar projects or inspire you to start making things
because it is easy, fun and creative. Subscribe to the channel now if you do not
want to miss any future video. A very promising new board hit the market
last year. It is the OpenMV M7 Camera module. The project started as a Kickstarter campaign
three years ago which managed to raise 100.000$ to bring the idea from paper to reality. The project aims to bring Machine Vision closer
to makers and hobbyists. They want to become the Arduino of Machine
Vision. I love this idea! It is going to make Machine Vision easy and
fun to use. Our projects will be able to “see” the
world for the first time! Let’s see what the board has to offer. The OpenMV M7 camera module features the following:
• OV7725 Image sensor with a resolution of 640×480 pixels
• ARM Cortex M7 processor • CPU Frequency: 216MHz
• 512KB RAM Memory • 2MB of FLASH memory
• microSD Card Slot • A full-speed USB interface to the computer
• SPI, I2C, CAN buses • 10 I/O pins
• A 12bit DAC and 12bit ADC • And, an RGB LED and two high power 850nm
IR LEDs. • MicroPython support
• Low power consumption, about 1W Compared to an Arduino Uno, or even an ESP32
we can see that this board uses a very powerful processor. The board costs around 50$, and you can find
a link to it in the description of the video below. OK, the specs are great but what can we achieve
with this board? The OpenMV Cam can be used for the following
things according to its designers: • Color Tracking
• Face Detection • Eye Tracking
• QR Code Detection/Decoding • AprilTag Tracking
• Line Detection • Image Capture
• Video Recording The list of possible applications is impressive. With the use of this board, we can move our
projects to the next level. The board can be programmed with an IDE developed
specifically for this board. The board runs the MicroPython operating system
so we can program the board using Python! How cool is that! But enough with the specs. Let’s build something with the board. At first, we have to download the OpenMV IDE. The software is free and easy to use. Each time you use it, the software asks for
a registration which costs 15$. Since I really love the idea and the board
I quickly registered to support the project. First of all, we have to connect the board
to the computer using a USB cable. Next, we have to press the connect button
and now we are ready to write our first program. Let’s write a simple blink project. We can program the board using Python! When the program is ready all we have to do
is to press the Start button and the program will execute, the LED will start blinking! But the program is not saved to the board
yet. In order to do so, we have to go to Tools
and select the “Save open script to Open MV cam”, press yes on the Pop-Up message
and we are done. Now if we unplug the board from the computer
and provide power from a power bank we can see that the blink program is running! We have successfully programmed the OpenMV
cam. If your program does not run if you unplug
it from the computer, be sure you have selected the correct Drive to associate with the board. If you don’t select the correct one, the
program won’t be saved. Now, that we know how to program the board
let’s look at another more advanced example. Jerry Lee, a viewer of the channel suggested
that I should take a look at AprilTags and openMV. So let’s load an AprilTag example. I was not aware of AprilTags. They are similar to QR codes but they encode
far smaller data payloads, between 4 and 12 bits allowing them to be detected more robustly
and from longer ranges. Further, they are designed for high localization
accuracy. You can compute the precise 3D position of
the AprilTag with respect to the camera. I simply printed an AprilTag on my printer. Let’s see if the OpenMV module can detect
it. I have loaded an example that comes with the
IDE. As you can see the OpenMV cam module can easily
detect the AprilTag. It can also detect its rotation and distance
from the camera! Amazing stuff! We can build some interesting projects with
this functionality. I think this functionality is very useful
to robotics. We can easily make this robot I built some
time ago to follow an AprilTag! Since I am a complete beginner to AprilTags
I have to learn more about them before using them, so stay tuned. While searching at the examples programs the
board offers, I came across Machine Learning examples! Can a 50$ board execute Machine Learning applications? It seems that it can! I loaded a demo program and a small pre-trained
model! The openMV camera module can detect cats,
dogs and other objects in real time! Let’s load an image of a dog. As you can see the program can detect the
dog with an accuracy of more than 70%! Of course, the accuracy is not great, but
the fact that a low-cost board can run machine learning models is a huge step forward! The possibilities are endless. The board needs around 160mA of current when
running the machine learning example! The power consumption is very low, way lower
than the power consumption of the Raspberry Pi which can achieve similar results. I am really impressed by this board. Artificial Intelligence is coming fast and
it is going to be the next big thing in the maker community. In a couple of years, boards like this one
will be common and less expensive. Our devices will be able to “see” the
environment using Machine Vision. Many problems can be solved if our devices
can see and understand their environment. Exciting times ahead! I can’t wait to build something with this
board! I would love to hear your opinion about the
OpenMV camera module. Do you like it, and are you going to use it
in any of your projects? Please post your comments below. Thanks! If this is your first time here, I would love
to have you subscribed. In this channel, I post videos about DIY projects
twice a month. I love making things, and I believe that anyone
can make things, anyone can become a maker. That’s why I created this channel, to share
my knowledge with the community and learn from the community. I hope you will join us. I will see you in the next video! [ Translating these subtitles? Add your name here! ]

77 comments

  1. Seems interesting but $50 for the board and $15 to register the software, is a bit steep for something I don't really have an application for (at this time). But I will keep this in mind if I ever run into a situation where I need computer vision, thanks.

  2. Thanks Nick the video was beautiful again. Unfortunately, it is something on the expensive side for me. But it is a nice board. maybe for the future if it gets cheaper.

  3. I am pretty sure the ESP32 runs at 240MHz Max, not 160MHz.
    I have a RPi3 + Rpi Camera already. USD $50+ price tag is a bit steep for a beginner, or what you get.
    OpenMV runs on all the Pi's since its in python.
    Then again, RPi is a power hog, and not battery friendly.

  4. Very neat if you have a low power budget for sure. No wifi and steep price will make it a tough sell for most though. Thanks for sharing

  5. It looks interesting, but as others have said it's a bit too pricey for a non application specific project IMO/on my budget. I imagine the software registration thing is due to the probability of clones hitting the market in the near future if the hardware sells well. I understand, no one want's to see their pet-project turn into a Saleae USB Logic analyzer like clone disaster. Still I expect full functionality out of the box for a 1 time purchase price when I buy something.
    Perhaps they should have done a hardware serial ID chip or something to control hardware clones. I have no idea how effective such a device is in practice. I've just skimmed a couple Maxim datasheets for their single wire interface versions.

    All that said, I'm not at all familiar with Python…yet.. I'm doing good to find my way around in C/C++ so I would probably struggle with this hardware.

    If your gaging interest on this subject in order to determine future content to create. I'm rather curious about Machine learning. I'll watch almost anything about the subject. Looking at how many views/subscribers YT CC's like the channel "Code Bullet" have (no affiliation), its quite a popular subject right now.

    Personally, I'm working on several projects and really really shouldn't add any more, but I would like to find a simple step by step guide for a practical/simple machine learning project using a STM32, ATSAM21, or any other hardware that can be sourced for under $10 and is available in a TQFP or larger pitched pin package (because they are within my current home etching capabilities).

    If your interested in FPGA stuff Sparkfun started carrying a new devboard based on a 32 pin QFN Lattice IC that apparently has a large open source project behind it. I'm curious about that one too altough I'm already playing with a CPLD and Altera Quartus so I shouldn't just abandon that.

    Thanks for the upload. I hadn't heard of or seen this camera project before. It's new to me 😉

    -Jake

  6. I see you have been watching video creators… good to see you using the tips well, I just need to do the same!

  7. Tempted to use it for watching a bird box or feeder next year and remove false triggers with the AI but adding WiFi and license pushes the cost way above the Pi solution I’ve looked at.

  8. " 640×480 8-bit Grayscale images or 320×240 16-bit RGB565" for $50 ?
    You can buy IP camera boards for around $10 (720P) to $20 (1080P)
    which output digital video to ethernet in a standard format.
    https://www.aliexpress.com/af/ip-camera-board.html?g=y&isNew=n&blanktest=0&origin=n&isFreeShip=y&jump=afs&SearchText=ip+camera+board&isBigSale=n&CatId=0&isFavorite=n&initiative_id=SB_20180917154611&needQuery=y
    You just need the software to interpret them.

  9. Great and well explained Edu hope you can do more projects with this Vision … Here is a robot I tried https://youtu.be/1enFgfFDQkM

  10. For the same $50, you can get an ESP32 (Yes it has 2 cores and can run at 240 MHz), and a camera that can support 1080p resolution and frame rate ( https://www.amazon.com/Arducam-Module-Camera-Arduino-Mega2560/dp/B013JUKZ48/ref=pd_bxgy_421_3?_encoding=UTF8&pd_rd_i=B013JUKZ48&pd_rd_r=848d3817-bb08-11e8-951a-bffc31d64799&pd_rd_w=XEndw&pd_rd_wg=57qkL&pf_rd_i=desktop-dp-sims&pf_rd_m=ATVPDKIKX0DER&pf_rd_p=6725dbd6-9917-451d-beba-16af7874e407&pf_rd_r=9AV36PV5N6V2J9FJ6AJZ&pf_rd_s=desktop-dp-sims&pf_rd_t=40701&psc=1&refRID=9AV36PV5N6V2J9FJ6AJZ)

  11. Can you make a tutorial how to make a TV Ambilight RGB with Arduino or Raspberry pi? There are some tutorials of how to make for monitors but It will be great if you make one of how to make for TV

  12. Πολύ ενδιαφέρον, θα σκεφτόμουν κάτι σε ταυτόχρονη αναγνώριση πολλαπλών προσώπων, σαν κάμερα ασφαλείας στην εξώπορτα που αν δει ένα "γνωστό" πρόσωπο, να κλείνει τον συναγερμό ή να σε ιδοποιι ότι ήρθε ο Τάκης π.χ.

  13. Great review. I've been working on a similar project. Can this one do face recognition or sample image matching? How about multiple blobs or edge segmentation?

  14. So expensive for what it is. Pi zero w, $15, and it has a 1GHz processor. With Pi cam (picked up cheap from China), it ends up way cheaper than this – around half price – AND Bluetooth and WiFi are built in.
    Interesting board though.

  15. I think $50 for the board is only a little bit to pricey (try $35) but $15 for the registration is about 5 x beyond way to expensive.
    Lower the board to $35 and then the lowered $2 registration fee will only be a little to expensive.

  16. Great video! Is there any chance you can do a video on displaying the camera onto a lcd to act like a home made reversing camera

  17. Did you ever notice your blink program blinks the GREEN LED and the documentation says LED 3 is the BLUE one.
    My board does the same as yours. also my LED 2 is RED when it should be GREEN. and LED 1 does nothing.
    Not sure about LED 4 since I can't see IR.

  18. There are mobile phones starting from $60. You know, those thingys with battery, cpu, 2 types of memory, display, wifi, 4g, gps, accelerometer, dust proof enclosure. Accidentally they have camera too… Actually, 2 cameras.

  19. Or you can get quad core 1ghz+ orange pi zero with full fledged linux and networking for like 9 USD. And whatever camera you want.

  20. H7 is now on kickstarter and soon will be on stores. It has faster processor and can use camera modules (like thermal) … I do believe though that banggood doesn't sell the original camera

  21. Hey, maybe you bought a plagiarized hardware in China. If you buy from the official website, this problem does not really happen. The kickstarter is really cheap! JUST $49 for H7. https://openmv.io/

  22. Machine vision market is valued at $9.89 Billion in 2017 and is expected to reach $19.95 Billion by 2024 at a CAGR of 8.10%.

    Request a sample @ https://www.envisioninteligence.com/industry-report/machine-vision-market/?utm_source=yt-chitti

  23. Machine Vision Market is valued at $9.89 Billion in 2017 and is expected to reach $19.95 Billion by 2024 at a CAGR of 8.10%.

    Request a sample @ https://www.envisioninteligence.com/industry-report/machine-vision-market/?utm_source=yt-hema

  24. Machine vision market is valued at $9.89 Billion in 2017 and is expected to reach $19.95 Billion by 2024 at a CAGR of 8.10%.

    Request a sample @ https://www.envisioninteligence.com/industry-report/machine-vision-market/?utm_source=yt-chitti

  25. Very cool thx for your videos. So fun ! About april tags, is it some lib in python, or you have write something to do that?

  26. Can I use openCV with this cam? I bought the pixy 2 but it doesnt return the raw image. Is it the same for this one?

Leave a Reply

Your email address will not be published. Required fields are marked *