dilipkumar25 / smartecam-ai-samples

A set of sample Deep Learning applications to get started with on the SmarteCAM.

SmarteCam Demo Applications

Deep learning application for people counting / crowd counting demo based on the detected heads.

Installing the demo applications

1. Download / clone the repository.

2. Run the script to install all dependencies for AI application.

   sh ./install.sh

3. Create directory structure:

   sudo mkdir -p /opt/econ/dlapp

4. Copy the application into SmarteCam device:

   sudo cp -r smartecam-ai-samples/ /opt/econ/dlapp

5. Download the people counting model file from Google Drive.

6. Download the crowd demo model file from Google Drive

7. Copy the model file to the models location:

   # people counting model
   sudo cp /path/to/trt_head_detection_FP16.pb /opt/econ/dlapp/smartecam-ai-samples/models
   
   # crowd demo model
   sudo cp /path/to/trt_nofpn_crowd_v1.pb /opt/econ/dlapp/smartecam-ai-samples/models

8. Launching the people counting application

9. Launching the crowd heatmap application

Deepstream Sample Appliations

Deepstream sample application for People Tracking

Executing the sample application

1. Open up a terminal

2. Run the config file using deepstream-app

deepstream-app -c /opt/nvidia/deepstream/deepstream-5.0/samples/configs/deepstream-app/source4_1080p_dec_infer-resnet_tracker_sgie_tiled_display_int8.txt

Preview

Prerequisites

• The above demo applications support v1.1 version of the SmarteCam.
• SmarteCam is connected to a 1080P HDMI monitor using a mini HDMI to HDMI cable.
• SmarteCam is connected to a PC using Ethernet cable.
• Demo application can be launched using a SSH session.
• Running both the application simultaneously is not recommended.