Output of MobileNetV2 SSD FPN-Lite 320x320

Help
#1

Question/Issue:
it is mentioned this block outputs the bounding boxes of the objects detected. Any input i give returns a (1,10) array always. The values are all floats. eg. [[0.02056512, 0.01845481, 0.01591653, 0.01008134, 0.00985926, 0.00638176,
0.00619035, 0.00511923, 0.00404363, 0.00337179]].

How do i interprete these?.. how can i get the bounding boxes?
Project ID:
185370
Context/Use case:
i have downloaded the .lite file after training. I loaded the model to custom code and made inference but could not interprete the output. Please help

#2

Hi @el_dAna,

The result from object detection (i.e. the result parameter provided to run_classifier()) should contain an array of ei_impulse_result_bounding_box_t structs. You can read about that struct here: ei_impulse_result_bounding_box_t - Edge Impulse API.

Here is an example showing how to read the bounding box information: example-standalone-inferencing/main.cpp at master · edgeimpulse/example-standalone-inferencing · GitHub

1 Like