linura/realtime_object_recognition

Realtime Object recognition using OpenCV 3.3 dnn module and pretrained MobileNetSSD caffe model

Realtime Obejct Recognition

Realtime object recognition using the OpenCV 3.3 dnn module + pretrained MobileNetSSD caffemodel.

Installation

All the dependencies can be installed using pip. Just use the following command from the root directory of the project.

pip3 install -r requirements.txt

NOTE: At the time of this writing, public.py is broken. If you have a problem installing it, check this workaround here.

How to run this script?

There are two options for video source:

• Webcam
• Android device running IP Camera (https://play.google.com/store/apps/details?id=com.pas.webcam&hl=en)

To run the script using webcam as source :

python3 real_time_object_detection.py --prototxt MobileNetSSD_deploy.prototxt.txt --model MobileNetSSD_deploy.caffemodel --source webcam

To run the script using IP Webcam as source, open the real_time_object_detection.py and edit the following line to match your host :

host = 'http://192.168.0.101:8080/'

Then to run the script using IP as source :

python3 real_time_object_detection.py --prototxt MobileNetSSD_deploy.prototxt.txt --model MobileNetSSD_deploy.caffemodel --source web

For any questions, create an issue in this repository.