Still Unable to Get Accurate Temps from Lepton3.5 Even after Firmware Update


I updated the firmware with the bin file you sent. It appears to update properly but at the end of programming the new firmware, It looks as though it is never completing the built in self test as the blue LED never blinks. I have waited over 30 min and it still is not flashing. Does this mean it is NOT programming the new firmware properly?

After about 30 min, I did click the “OK” button and the H7 board appears to be working but it still does not display accurate temperatures. It is doing the same as before so I wonder if for some reason the firmware is not programming correctly. I have attached a picture with several ROIs. The temperature in the ROIs should be around 45 degrees C per a temperature probe I have on hand.

This picture was taken after about 2 hours of warm up on the Lepton 3.5. I am not sure what is going on. You also mentioned the possibility of using the spot-o-meter function to get the temperatures. Can you please provide some sample code for using this function?

Thanks in advance for your help!

Ken Taylor

If you didn’t erase the file system the script onboard the camera will continue to run. That’s why the blue led didn’t blink. Should add an if statement on the ide text to handle that…

Um, regarding the temp. Let me check out what’s going on.

Hey, can you give me a raw image capture with markups of what you are capturing? Also, please make sure you are running the target temperature capture script that puts the Lepton into the mode where you get the actual temperature value from the image versus AGC mode (the default) that just makes up pretty images.

With a raw GRAYSCALE image and I see if there’s anything wrong. Please edit the script such that the image is not colored.

I am just using one of the example scripts (lepton_get_object_temp_3) to make sure I have not introduced any errors. I think I partially figured out what is going on. I think the problem is when I set the measurement range…specifically the max temp it is scaling the results in the ROIs which I think it normal. As I increase the the max temp, the temp in the ROI gets closer to the actual temperature.

Should this script also find ROIs that are cold (appx 10 degrees C) providing the pixel size and area are correct? I am only having success seeing HOT objects. Would the spot-o-meter code you mentioned work better for this?

Also, please tell me how to save a GRAYSCALE image to send to you. I know how to save an image but not sure how to make sure it is GRAYSCALE.


Okay, I see your issue.

So, you have to set the min and max temp range. The whole image is filled with pixels where the min and max temps represent 0 to 255. That said, as you increase the temp range it’s harder to see subtle temperature contrasts because the image is scaled across such a wide range.

Then, for finding the min temp objects, the code is written to only look for pixels values near 255. Do you see the color thresholds in the script? If you change those thresholds to something else like values near zero the find_blobs() algorithm will find the min temp parts. Note that can can call find blobs twice with two different sets of thresholds to find min and max temp objects.

To be clear, the script just finds one temp range right now. If you want more you need to copy and paste the code that does stuff with find_blobs() to run twice.

Anyway, just keep in mind that the whole picture is just pixels with values of 0 to 255 where 0 represents the min temperature you set and 255 is the max temperature. Everything in between is just a linear scale.

You probably don’t need to save an image. If you aren’t setting the min and max right the results you saw would be expected.