Face recognition using dlib with Jetson nano

shubham.shah09 · January 30, 2021, 4:53am

Hi,
I am using the following tutorial to implement face recognition on Jetson nano
post link

The tutorial uses Dlib and face_recognition library to do the task of face detection and recognition. it fetches 6-7 FPS on Jetson nano, how can I incorporate this tutorial with deepstream as i want to run it for several camera streams.

thanks

AastaLLL · February 1, 2021, 3:08am

Hi,

dlib has its own detection and recognition model.
If you can convert the model into TensorRT, you can just replace the path and parser in a standard pipeline:

If the model doesn’t be fully supported, you can try to insert the dlib implementation into the below sample.

github.com/NVIDIA-AI-IOT/deepstream_python_apps

apps/deepstream-test1-usbcam/deepstream_test_1_usb.py

master

#!/usr/bin/env python3

################################################################################
# SPDX-FileCopyrightText: Copyright (c) 2020-2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: Apache-2.0
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
################################################################################

import sys

This file has been truncated. show original

Please noted that the buffer transfer between components is a GPU buffer.
Please check if dlib can accept the GPU buffer or not first.

Below is the TensorRT support matrix for your reference:

Thanks.

shubham.shah09 · February 1, 2021, 3:15am

Thanks @AastaLLL, I’ll check all the methods you mentioned and get back. Thanks once again

shubham.shah09 · February 2, 2021, 10:51am

is there any already existing pipeline for face recognition with deepstream?
thanks

shubham.shah09 · February 2, 2021, 11:04am

Is there a way to just get a video feed with deepstream from multiple sources, I don’t want to run any inference.

AastaLLL · February 3, 2021, 8:39am

Hi,

We don’t have a face recognition pipeline since there is no face related model in the Deepstream.

To feed multiple sources into Deepstream is easy.
For example, with /opt/nvidia/deepstream/deepstream-5.0/samples/configs/deepstream-app/source30_1080p_dec_infer-resnet_tiled_display_int8.txt.

1. Turn off inference

...
[primary-gie]
enable=0
gpu-id=0
...

2. Run

$ deepstream-app -c source30_1080p_dec_infer-resnet_tiled_display_int8.txt

Thanks.

shubham.shah09 · February 3, 2021, 10:34am

Awesome, thanks. how can i do the same in python @AastaLLL

shubham.shah09 · February 3, 2021, 10:41am

where can i access the frames of this stream @AastaLLL

AastaLLL · February 5, 2021, 4:44am

Hi,

You can find some python-based sample in our GitHub below:

To access the image data, please check the following comment for information:

Thanks.

shubham.shah09 · February 5, 2021, 6:01am

Thanks @AastaLLL I’ll look into it.

shubham.shah09 · February 5, 2021, 12:25pm

how can i disable pgie and tracker in the following example
thanks

AastaLLL · February 9, 2021, 7:15am

Hi,

You can comment out this line:

github.com

NVIDIA-AI-IOT/deepstream_python_apps/blob/master/apps/deepstream-imagedata-multistream/deepstream_imagedata-multistream.py#L355


      
          print("Creating EGLSink \n")
          sink = Gst.ElementFactory.make("nveglglessink", "nvvideo-renderer")
          if not sink:
              sys.stderr.write(" Unable to create egl sink \n")
          
          
if is_live:
              print("Atleast one of the sources is live")
              streammux.set_property('live-source', 1)
          
          
streammux.set_property('width', 1920)
          streammux.set_property('height', 1080)
          streammux.set_property('batch-size', number_sources)
          streammux.set_property('batched-push-timeout', 4000000)
          pgie.set_property('config-file-path', "dstest_imagedata_config.txt")
          pgie_batch_size = pgie.get_property("batch-size")
          if (pgie_batch_size != number_sources):
              print("WARNING: Overriding infer-config batch-size", pgie_batch_size, " with number of sources ",
                    number_sources, " \n")
              pgie.set_property("batch-size", number_sources)
          tiler_rows = int(math.sqrt(number_sources))
          tiler_columns = int(math.ceil((1.0 * number_sources) / tiler_rows))

Thanks.

shubham.shah09 · February 9, 2021, 8:05am

Thanks @AastaLLL

Topic		Replies	Views
How to use deepstream to detect and capture faces DeepStream SDK	4	795	October 12, 2021
Deepstream sample to usefor our solution Jetson Nano	6	402	October 14, 2021
Face Recognition Sample in DeepStream DeepStream SDK	11	1129	October 12, 2021
Deepstream support for Face detction DeepStream SDK tensorrt , jetson-inference	5	1919	October 12, 2021
Face Recognition Deepstream App DeepStream SDK	7	1445	October 12, 2021
Consultancy Needed for a quick POC DeepStream SDK	2	532	October 12, 2021
Hardware Accelerators Jetson Nano Camera to Face recognition Jetson Nano gstreamer	2	875	October 18, 2021
DeepStream for MTCNN and FaceNet DeepStream SDK	4	1439	October 12, 2021
Facelandmark detection in jetsons with python DeepStream SDK	2	925	October 12, 2021
Regarding doubts about face recognition in deepstream DeepStream SDK gstreamer	2	468	June 19, 2023

Face recognition using dlib with Jetson nano

Related topics