BACK-TO-BACK-DETECTORS REFERENCE APP USING DEEPSTREAMSDK 4.0
The project contains Back to Back detector application to show the capability of Deepstream SDK.
Please follow instructions in the
apps/sample_apps/deepstream-app/README on how
to install the prequisites for Deepstream SDK, the DeepStream SDK itself and the
Export the environment variable:
export DS_SDK_ROOT="your deepstream SDK root"
Preferably clone the app in
Steps to download the models:
- To download the models for the second nvinfer, visit: https://github.com/NVIDIA-AI-IOT/redaction_with_deepstream
- Use the following commands:
$ cd $DS_SDK_ROOT/samples/models $ mkdir Secondary_FaceDetect $ cd Secondary_FaceDetect $ wget https://github.com/NVIDIA-AI-IOT/redaction_with_deepstream/raw/master/fd_lpd_model/fd_lpd.caffemodel $ wget https://raw.githubusercontent.com/NVIDIA-AI-IOT/redaction_with_deepstream/master/fd_lpd_model/fd_lpd.prototxt $ wget https://raw.githubusercontent.com/NVIDIA-AI-IOT/redaction_with_deepstream/master/fd_lpd_model/labels.txt
$ make $ ./back-to-back-detectors <h264_elementary_stream>
- Edit the paths in
secondary_detector_config.txtto the location of the models downloaded from the above site.
This document shall describe about the sample back-to-back-detectors application.
This sample builds on top of the deepstream-test1 sample to demonstrate how to add multiple back-to-back detectors in the pipeline.
Two instances of the "nvinfer" element are added to the pipeline serially after nvstreammux and before the display components. Both the "nvinfer" instances have their own config files.
The first "nvinfer" instance (Person/Vehicle/Bicycle/RoadSign) will always act as primary detector.
The second "nvinfer" instance (Face/License Plate) can be configured as
primary(full-frame) / secondary (operating on primary detected objects). By
default it is configured in the secondary mode. To change the second "nvinfer"
instance to primary mode, change the macro
SECOND_DETECTOR_IS_SECONDARY in the
sources to 0.