Wrong prediction every time using the person detection model on ESP32-WROVER-IE board

All ESP32 boards running MicroPython.
Target audience: MicroPython users with an ESP32 board.
Post Reply
Aduka_27
Posts: 27
Joined: Thu Apr 14, 2022 11:00 am

Wrong prediction every time using the person detection model on ESP32-WROVER-IE board

Post by Aduka_27 » Fri Apr 29, 2022 12:02 pm

Hi guys,

I am using the ESP32-WROVER-IE board with the ov2640 camera connected to it. I am trying to run the person detection example (https://github.com/mocleiri/tensorflow- ... n-examples) and see if I can get some output results based on the captured image. Every time I get same result where the higher value is assigned to not_a_person prediction even if I am present on the picture. I also tried with the existing example but there as well the prediction is always negative (I mean it's always like there is nobody on the picture). Is there any possible solution for this issue? Any kind of help is welcome.

Post Reply