security barrier camera demo cannot detect my images of cars

[Lee__Ken]

I've searched but cannot find any posts that is related to my issue.

The "security barrier camera" demo does not work when i tried to use my own images of cars (in jpeg), even though it works fine with the default image file "car_1.bmp".

I'm just starting with OpenVino. Have successfully setup and run both  demos under the folder of "deployment_tools\demo":

"squeezenet download convert run" and "security barrier camera"

It works fine with my own images for "squeezenet download convert run".

It does not work, however, when I tried to use my own images of cars with the demo of "security barrier camera". It opens the image at the end, but without any text for the results of detection of the color and license plate. The demo does works fine, however, with the default mage of "car_1.bmp" in the "demo" folder.

I also tried to convert my own images(in RGB) to BGR format, but still with no luck.

The example image "car_1.bmp" has a license plate from China, does that mean the demo app only works with Chinese license plates?

or the size of the images have to be in BMP format? or it has to be in certain sizes?

thanks 

[Monique_J_Intel]

Hi Ken,

Yes the model was trained on all license plate images from China. To understand more of the model's functionality you can go to our open model zoo github where every model has a mark down file that will tell you relevant information on how it was trained, metrics upon training, etc. 

Let me know if you have any other questions pertaining to your post.

Kind Regards,

Monique Jones

[Lee__Ken]

Hi Monique,

Thanks for the info. I will review the model zoo.

I noticed that you are from Intel. Could you please pass on my comments on OpenVino? It would be great helps for beginners to include more media files in the OpenVino deployment package, for example, for the "security barrier camera" demo, currently just one image file only: "car-1.bmp". The image files usually won't take much spaces, so the more the better.

Also, it will also be great if including more details regarding the input parameters, especially on the requirements of images files, for the demos and samples.

thanks

Ken

[Vladimir_Dudnik]

Hi Ken,

thanks for your feedback, we definitely will take that into consideration.

Regards,
  Vladimir