Poor performance on SSD example versus security barrier camera demo

[goh__richard]

We have trained our own SSD object detector model

It is able to detect well using the security barrier camera demo sample program.

We are able to input both full hd images and old 768x576 resolution images.

Both resolutions are detected well.

When using the object detection ssd async sample program with the same trained model. (xml and bin)

The program is able to detect the full hd images ( 1920x1080)

but is unable to detect any objects for the smaller 768x576 resolution images.

How can we make the ssd async example detect as well the smaller 768x576 images?

Appreciate any advise.

[Iffa_Intel]

Greetings,

Both image resolution works for me and video also works.

The trick is to CHANGE the wait_key_time = 1 TO wait_key_time = 0 in object_detection_demo_ssd_async.py.

Attached are my results. Note that you need to click x to close instead ctrl c.

Sincerely,

Iffa

[Iffa_Intel]

Greetings,

Intel will no longer monitor this thread since we have provided a solution. If you need any additional information from Intel, please submit a new question.

Sincerely,

Iffa