Efficientnet-b0

You can read more about the EfficientDet model in OpenVINO's Docs This model includes 1000 classes of image classification. You can view these classes in the classes.py script.

Instructions on how to compile/Install the model yourself:

blobconverter -zn efficentnet-b0 -sh 6

Specify the number of Shaves to install accordingly. This command requires Blobconverter v0.10.0

Demo

The demo classifies the animals in the images as Ibex (Mountain Goat) accurately. Also classifies most common objects.

Installation

python3 -m pip install -r requirements.txt

Usage

Run the application

Use the DepthAI 4K RGB Cam Input Feed:

python3 main.py -cam

Use any of your own MP4 videos as input. Video resolution of 224x224 is recommended for better visualisation output.

python3 main.py -vid [vid path]

Use the -nd command to not display RGB demo and instead print data to the console

python3 main.py -nd