For more projects - kzitem.info That Project Github Repository - github.com/0015/ThatProject Join FB Group - facebook.com/groups/138965931539175
@GadgetAddict
2 жыл бұрын
That's pretty incredible. I never would have thought you could get opencv working on something with such little processing power.
@ThatProject
2 жыл бұрын
I thought so too. It's slow but works well, so I think it can be used in many projects.
@dennis8196
2 жыл бұрын
There is a lot of power in those ESP32's if used correctly.
@THEbonny95
2 жыл бұрын
I asked myself this question not too long ago. Why can't opencv run on esp32. And today, you did it. =)
@gustavodullens3107
7 ай бұрын
Hello! Im trying to implement this on the ESP32-CAM (Ai Thinker)! I was successful while using cvtColor and threshold and imwrite, but the moment I tried the Canny function i got the sysconf error. Tried following the guide but while building the library on the docker it returns this "/bin/bash^M: bad interpreter: No such file or directory". Can you help me?
@ThatProject
7 ай бұрын
When I tried to build Canny Edge Detection using Docker, there were no problems. Are you sure there are no problems using other functions? Since this is included in imgproc, there should be no problem as long as it is built.
@codevanguard
10 ай бұрын
this project is the answer why expert do not use Arduino IDE but using ESP-IDF ESP-IDF DEV Environment cool
@fenderrexfender
2 жыл бұрын
makes me want to work on my cars camera project.
@kevindarren756
7 ай бұрын
Thank you for the extremely informative video, but i have some questions of my own: 1. Which esp idf version did you use with this, im using 5.2v and getting some warnings when building the files, regarding a wrong type arithmetic operation. 2. Before even trying this, basic cv::imread, does not work for me and throws out the same sysconf error. Is there a fix to correct this for any project i want to run?. I've tried to do the docker edits and copy paste the file from ttgo_demo to my project and replace the opencv directory there, did not work. 3. I have alot of warnings when building any of the files, all from the opencv directory inside main?, any clue what it could be? Im running it on Ubuntu 22.04, Any help would be appreciated!
@ThatProject
7 ай бұрын
kzitem.info/news/bejne/mKeGr4Z4kmh0a5gsi=9CfiSoThovj8_rlk&t=371 As you can see here, I built it using Docker Container. The reason is that I could not fix the error that occurred when building in my environment. Anyway, the IDF version of this Container is 4.2, so if you are using the latest IDF 5.x you will need to build ESP32-Opnecv in your environment. I tried it too, but after a few failed attempts, I'm not wasting any more time. For now, I think it's best to build using this container until there is a new update.
@markuspfaff5504
Жыл бұрын
hi, great video. Have you ever tried the serial monitor with the ESP32-CAM ? I using the ESP32-CAM with the programming board (MB). No chance to get output text in serial monitor. If you have no problem with the serial monitor, waht have you changed? :-)
@ThatProject
Жыл бұрын
Do you use FTDI Programmer when flashing to ESP32-CAM? I think there is no problem with serial output through this. Are you using Arduino? or ESP-IDF?
@adityasaiakella1287
11 ай бұрын
okay i tried building and uploading this code on my AI-Thinker ESP32 Cam board through vscode, now is there any way to see the images on which the opencv works, or is it compolsury to have a display. let me explain, the camera web server code helps us stream through the cam in the AI Thinker board, like that is there any way that a stream opens for this and test the features of opencv like say object detection or anything?
@ThatProject
11 ай бұрын
The purpose of this project is to operate opencv as a standalone based on ESP32. That's why I think a display is essential. In general, it is common to use ESP32CAM to transmit acquired images to a server and perform image processing on the server. Examples of this can be found by Googling.
@brahimlawal
Жыл бұрын
Need help making my esp work as a stand-alone for license plate recognition Any suggestions?
@edgarmondragon4708
2 жыл бұрын
Did you try this on the esp32s3? I suppose that to take advantage of the esp32s3 vector instructions it is necessary to rewrite some code...
@ThatProject
2 жыл бұрын
No, I haven't had a chance to do that yet. Perhaps a lot of things need to be modified.
@naiwww1857
2 жыл бұрын
it supposed to be work better than this tho, especially one who have riscv architecture
@mairswartz
2 жыл бұрын
That's very cool. Thank you for sharing
@anujsreenivasan5040
3 ай бұрын
Bro can you detect aruco markers using this? I have the repo cloned and it doesn’t seem like it
@ThatProject
3 ай бұрын
This build of openCV does not include the ArUco library.
@anujsreenivasan5040
3 ай бұрын
@@ThatProject ok thanks
@PhillipHilton
2 жыл бұрын
Awesome man!
@muhammadsalem9924
Жыл бұрын
I want to implement this project: "Design of Driver Awareness system depending on Eye and Face Feature and Discrimination" Can I use ESP32 instead of Raspberry Pi ? How do I program it and what do you use ? please reply
@ThatProject
Жыл бұрын
Absolutely not recommended. Running OpenCV independently on ESP32 is possible, but very slow. The application you are trying must run on SBC, not MCU.
@muhammadsalem9924
Жыл бұрын
@@ThatProject Thank you so much Please, what are the Raspberry pi specifications required to implement the project?
@ThatProject
Жыл бұрын
I believe that RPI4 is good enough for your project.
@ionymous6733
2 жыл бұрын
I am considering this hardware for my project. I want to use it for something like circle detection. You had to limit it to 240x240 and I don't think that would be enough for me. Could I sacrifice FPS in order to increase camera resolution? For example, could it process the ov2640's full 1600x1200, even if that's more than 33 times more data, at something like 1 frame every 8 seconds? Or maybe memory or something else would still be an issue?
@ThatProject
2 жыл бұрын
That's a tricky question. It is true that on the ESP32 OV2640 you can get a resolution image of FRAMESIZE_UXGA (1600 x 1200). But displaying this on the screen can be a whole other matter. It can update the screen once every 8 seconds, but it could cause a memory overflow because of the size of the image object in LVGL.
@ionymous6733
2 жыл бұрын
@@ThatProject For my purposes, I would not need a screen at all. The results of any edge or shape detection would be used in processing, like controlling a robot. But anyway, you are saying there a memory issue. Is 240x240 already getting close to a memory limit in your edge detection demo? Could you make an educated guess at what the resolution limit would be in that app if you didn't care about displaying?
@ThatProject
2 жыл бұрын
@@ionymous6733 First of all, the biggest problem is that when it acquires the image, it gets the image in RGB565 type, not in JPEG image format. The resolution you want is only available in the JPEG image format. In the case of RGB565, the system does not work because it exceeds the available frame buffer size. In order to find out the possible resolution, it seems to be necessary to first modify the image so that it can obtain an image in JPEG rather than RGB565, and also modify the OpenCV code.
@RixtronixLAB
2 жыл бұрын
Creative video, thanks for sharing, like it :)
@sltechgalaxy1677
2 жыл бұрын
can we upload any OpenCV code to the esp 32 camera module using this method??
@ThatProject
2 жыл бұрын
Do you mean the ESP32 AI-Thinker model? Well, I haven't tested yet. If there is enough memory space, you can upload the OpenCV code as I did.
@sltechgalaxy1677
2 жыл бұрын
yes I mean the ESP32 AI-Thinker model, ok I will try it thanks for your reply
@nokimo9138
2 жыл бұрын
@@sltechgalaxy1677 I need this same, have you tried it yet? I probably will try use this in next days, i need some liblary with based simple image editing functions and I found this :D
@sltechgalaxy1677
2 жыл бұрын
@@nokimo9138 yes i also try but this week i can't test it because of a reason
@sltechgalaxy1677
2 жыл бұрын
@@nokimo9138 please update me also about your test
@mracevedo85
2 жыл бұрын
Is it possible to perform the hough transform to detect lines? Thank you
@ThatProject
2 жыл бұрын
I believe so. It's possible to find a specific shape using the hough transform with this system.
@Karslo
2 жыл бұрын
You're amazing!!! Do you think this could run on a LILYGO TTGO T-Camera v1.6.2?
@ThatProject
2 жыл бұрын
If you can use LVGL with an ssd1306, it might be possible, but I don't recommend it.
@UsmanKhan-l8l7k
6 ай бұрын
what about esp32 cam?@@ThatProject
@deanstarman1694
2 жыл бұрын
What! How can it be possible?? I don't get it.
@UsmanKhan-l8l7k
6 ай бұрын
will this work on esp32 cam?
@ThatProject
6 ай бұрын
You can also use it with ESP32-CAM, but as you know, it doesn't have a display. What calculations are you trying to make internally with OpenCV?
@QuadDerrick
2 жыл бұрын
when i will see something fun on this esp32 , like a webserver statistic display ? file server statistic ?
@ThatProject
2 жыл бұрын
Haha, I'll make it when I get a chance. The problem is there are too many interesting projects.
@mairswartz
2 жыл бұрын
@@ThatProject perhaps this is useful kzitem.info/news/bejne/s2-kmWaksaSrqJw
@gudimetlakowshik3617
Жыл бұрын
Hi, I need your help. I'm trying something similar and constantly getting error saying mutex not in namespace cv when using arm-none-eabi-gcc compiler. Can you please let me know your contact so I can post full details??
@ThatProject
Жыл бұрын
Unfortunately, I don't know anything about this. How about posting your issue here? github.com/joachimBurket/esp32-opencv
@Duloid77
Жыл бұрын
permission mister, i ask question for you, can input sistem android to arduino, to make smartphone, sorry, question out to topick you video
@ThatProject
Жыл бұрын
I'm not sure what you're talking about. Please give a more detailed explanation.
@guilhermemarins8807
2 жыл бұрын
does this work in a common esp32cam module?
@ThatProject
2 жыл бұрын
I think you can build it for ESP32-CAM(AI THINKER) but what is your plan without display?
@EderANavarroPereyra
Жыл бұрын
@@ThatProject I'm triying to show the processed image on a webpage, the problem is that I can't find the way to convert the opencv MAT into a camera framebuffer to send over http, can you help me?
@weirdsciencetv4999
7 ай бұрын
Can you do simple haar cascades?
@ThatProject
7 ай бұрын
As far as I know, this only supports core, imgproc, and imgcodecs. The problem is that you must have objdetect to use Haar-cascade Detection and it doesn't seem easy to include this.
@weirdsciencetv4999
7 ай бұрын
@@ThatProject yes I suspected this, but I have seen some hacking miracles before lol! Edit: also, great channel I subscribed!
@bennguyen1313
2 жыл бұрын
The D0WDQ6 datasheet seems to say it doesn't have any flash?! Assuming it has enough memory, do you think this build process could work on ESP32s that are based on Risc-V (ex. ESP32-C3FH4), instead of the Xtensa LX6? Any thoughts on using Visual Studio Code with the PlatformIO extension, instead of ESP-IDF extension?
@ThatProject
2 жыл бұрын
OpenCV can work based on RISC-V, but I don't know if it can run on ESP32 RISC-V yet. I'd love to try it too. github.com/opencv/opencv/wiki/OpenCV-RISC-V You can use Visual Studio Code with the PlatformIO extension instead of the ESP-IDF extension, but the build requires ESP-IDF.
@THEbonny95
2 жыл бұрын
What is this screen module?
@ThatProject
2 жыл бұрын
This is LILYGO TTGO T-Camera Plus. kzitem.info/news/bejne/yq6qumuhrWaKl34
@ele_mac
2 жыл бұрын
thanks, it's nice
@THEbonny95
2 жыл бұрын
Is it possible to detect an object by color?
@ThatProject
2 жыл бұрын
Well, usually we do detect an object by shape though. What do you mean "by color"??
@THEbonny95
2 жыл бұрын
@@ThatProject Well, maybe so. Will the controller work?
@jlcamry
2 жыл бұрын
very nice is it possible to convert this code for arduino ide?
@ThatProject
2 жыл бұрын
Unfortunately, I haven't been able to find a way to make it work from the Arduino base yet.
@hajarmiskar3305
2 жыл бұрын
Can esp detects custumized objects ?
@ThatProject
2 жыл бұрын
I'm currently testing various things. What specifically do you want to detect?
@hajarmiskar3305
2 жыл бұрын
Snake & scorpions as one class called danger
@ThatProject
2 жыл бұрын
@@hajarmiskar3305 That kind of detection requires using ML, and it's difficult to use with ESP32 OpenCV yet. Please take a look at TensorFlow Lite
@musaesin5163
2 жыл бұрын
Great
@vinacheapcoltd413
Жыл бұрын
Help on windows?
@ThatProject
Жыл бұрын
If you build using Docker, Windows should work just fine. Any problems?
@shanno88
2 жыл бұрын
pls check whether mediapipe is compatable?
@seanyong1123
2 жыл бұрын
I don't think it's powerful enough
@ThatProject
2 жыл бұрын
It seems impossible with ESP32.....
@chrisw1462
2 жыл бұрын
Pretty awesome work! If I may, please talk just a little slower? You have a pretty strong accent and it's hard to understand, especially the times when you talk _really_ fast.
@ThatProject
2 жыл бұрын
Thanks for your comment. I'll try to make it better next time.
@imcrazy886
2 жыл бұрын
The talk speed is ok for me.(English is not my native language.) Chris, You can force the playback speed(.25x .5x .75x 1x 1.25x 1.5x 1.75x 2x) in KZitem setting by yourself.
@ddegn
2 жыл бұрын
Why disable likes/dislikes? It looks like you have really interesting projects on your channel but I don't trust channels which disable likes/dislikes. I wonder if you also delete any negative comments to hide the fact your tutorials are bad. I'm not saying your tutorials are bad, it's just I'm more suspicious of your channel than channels with likes/dislike enabled.
@ThatProject
2 жыл бұрын
I never thought deeply about it. If enabling like/dislike can increase the credibility of my channel, I should enable it. Anyway, thank you. (Now you can check the like/dislike of this video.)
Пікірлер: 92