ESP32 CAM error on image_util.h

Hello all,

I am trying out edgeimpulse on a ESP32-CAM.

I Copied the code from here: https://github.com/edgeimpulse/example-esp32-cam/blob/main/Examples/Basic-Image-Classification/Basic-Image-Classification.ino

But, when running the code i get this error message show up:

Basic-Image-Classification:12:10: fatal error: image_util.h: No such file or directory

I don’t know where i can get that file or what i have to change.

I am just started with edge impulse, so this is my first try.

Thanks for the help!

More information:
Looks like this is the same issue: https://github.com/edgeimpulse/example-esp32-cam/issues/7

Hello @MarcVer,

Which version of the boards have you installed in your Arduino IDE?
Here is what I have on my side:

When I select the AI thinker board, I can compile successfully:

Also, image_util is present in my Arduino’s application folder:

Let me know if you solve your issue so I can update or improve the Github’s README.
If not I’ll try a fresh installation of the ESP32 boards.

Regards,

Louis

First of all, thanks for your help!

I have version 2.0.2 installed in boards manager.

In my Arduino15/packages/esp32/hardware/esp32/2.0.2/tools/sdk folder i have:
esp32
esp32c3
esp32s2

But no include folder.

I will try to downgrade my esp32 boards

@louis

Thanks! First one fixed :slight_smile:

Now i get this error from the edgeimpulse library.

‘struct ei_impulse_result_t’ has no member named ‘classification’

I think this is the same issue now: Compilation Failures running Object Detection Model on Arduino IDE for ESP32CAM board

But, is there a fix already?

Indeed, at the moment we cannot run Object Detection on most micro-controllers targets. If you have a look at the resources needed in your “Object Detection” tab in the studio, you’ll see that it does not fit on the ESP32 resources.

Two good news :tada: :

  • We have been working on a lighter object detection technique, it should be release very soon. Stay tuned :slight_smile:
  • We will officially support ESP32 in the following month.

Regards,

Louis

1 Like

@louis
Thanks!

I see your name now in collaborators in my project, is that ok?

I will follow the news en stay tuned for the esp32 release! Really looking forward to it :slight_smile:

@MarcVer, yes I joined your project to make sure you were using an Object Detection project before replying.

I’ll remove myself now from your project.

Regards,

Louis

No problem! Thanks for the great help! Will try again next month :slight_smile:

1 Like

@louis

One more question, does image classification work on the esp32 already?

In other words, is it only the object detection that is not yet supported?

Thanks again!

Hello @MarcVer,

Yes Image Classification works already on ESP32 although it is not officially supported and you’ll need to use a small MobileNet model (MobileNetV1 96x96 0.1) to stay within the ESP32 resources.

Regards,

Louis

@louis Thanks again! Will try that!

@louis

Are there any updates for the esp32 CAM object detection support?

Hello @MarcVer,

Not yet, we are still finalizing our constraint object detection models internally.

Regards,

Louis

1 Like

We are also working on full ESP32 support soon!

2 Likes

i have the same error message!
In the Arduino IDE and vscode platformio. I reinstalled the esp32 board on Arduino IDE but it didn’t help. Can someone give me a link so i can download the header file?
Thanks in advance!

Good news :slightly_smiling_face: :metal: it will be interesting to see

Hello @OEWPS, @norik.badalyan,

You can now use the example provided when you download the Arduino Library.

See Arduino library - Edge Impulse Documentation

Best,

Louis

1 Like

Hi @louis

I have followed each and every step mentioned but I’m getting the below errors. Could you look into it.

Arduino15\packages\esp32\hardware\esp32\1.0.6/tools/sdk/include/newlib/math.h:285:15: note: ‘fmin’
extern double fmin _PARAMS((double, double));

Hello @ravi1706,

This repository is now deprecated: https://github.com/edgeimpulse/example-esp32-cam.

ESP32 examples are now directly available in the Arduino Library deployments.

Best,

Louis