Birds Eye View on NVIDIA Jetson Boards

A Bird's Eye View (BEV) is an elevated view of an object from above, with a perspective as though the observer were a bird. It is often used in the making of blueprints, floor plans, maps and car parking systems. The result is obtained with a technique that generates an aerial view of a scene based on various lateral perspectives. Several cameras are used to obtain those lateral views, usually at least 4, to get a 360 view around an object. 

The project experiments with the notion of Bird’s Eye in Desktop PC and NVIDIA Jetson platforms.

To get a good field of view with the minimum amount of cameras, for example 180 degrees per camera, the image will be generated with a fisheye perspective. The RidgeRun’s BEV engine is able to remove the input distortion and using a preconfigured settings file generate an output image. This may be appreciated on Figure 1.

Figure 1. Example of a BEV image generated from 4 lateral fisheye cameras.

The process is composed of 6 smaller steps. First, the buffers are obtained from the camera. Next the fisheye distortion is removed. The third and fourth steps generate the elevated perspective with the technique known as IPM (Inverse Perspective Mapping). Fifth and sixth steps adjust the image to obtain the desired output section for each camera by applying an enlargement to match the perspective with the other cameras and by cropping the relevant section of the image. A summary of the process is depicted in Figure 2.

Figure 2. Simplified process to generate the BEV

The C++ library was developed in a modular way, open to include new image processing frameworks, but was initially tested with the help of OpenCV.

Version 0.2.0 includes:

• Fisheye and rectilinear test data sets (synthetic and real)
• Jupyter notebooks with detailed prototypes
• Fisheye correction and BEV computation
• C++ library with OpenCV based implementation
• CUDA acceleration
• Extensive set of examples

Some areas under development:

• Image Stitching integration
• GStreamer support

Learn more in our developers wiki or check out the API reference.

Contact us to be part of this research!

NVIDIA Jetson TX2-Camera image capture latency measurement techniques.

For better understanding of this latest blog from RidgeRun, please read the RidgeRun blog on Diving deep into NVIDIA Jetson TX2 - Video Input system and camera image capture latency.

The techniques used to measure the Jetson TX2 image capture latency are outlined in this blog.

Two timestamps are used to measure the latency :

SOF : The first timestamp is taken when the pixels of a frame start to arrive at the VI (t0), this is the start-of-frame (SOF).
The second timestamp is usually obtained at the time when the frame is available on a userspace application (t1).

Techniques can be divided into two main categories, techniques that use the CHANSEL_PXL_SOF (referred to as SOF) and the techniques that do not use the SOF .

Outline of both the methods:

Techniques that use the SOF timestamp

CHANSEL_PXL_SOF can be used as t0 and then, t1 can be the timestamp obtained at the instant when the frame arrives to userspace using ideally the same clock used to set the CHANSEL_PXL_SOF, then the latency can be computed as follows:
Latency = t1 - t0

Techniques that do not use the SOF timestamp

LED test
This test involves the use of a led connected to the TX2 GPIO.
The camera must start capturing when the led is still off, and then a kernel module must turn the led on and then record the current CLOCK_MONOTONIC timestamp to be used as t0. Each image that arrives to userspace must be timestamped also with the CLOCK_MONOTONIC at the instant when it's available, this will be t1.

More technical details about the techniques, working with TSC (Time Stamping System Clock) for the latency measurements and rtcpu, v4l2 tracing details are explained in the RidgeRun developer wiki page : NVIDIA Jetson TX2 - VI Latency Measurement Techniques

Contact Us

Please visit our Main Website for the RidgeRun online store or Contact Us for pricing information of the engineering support, product and Professional Services.
Please email to support@ridgerun.com for technical questions and for an evaluation version (if available).
Contact details for sponsoring the RidgeRun GStreamer projects are available at Sponsor Projects page.

NVIDIA Jetson Xavier Multi-Camera Artificial Intelligence Demo from RidgeRun

Jetson Xavier Multi-Camera Artificial Intelligence Demo from RidgeRun

This demo from RidgeRun shows the capabilities of the Jetson Xavier by performing :

• Multi-camera capture through FPD-LINK III with Virtual Channels support,
• Display of each individual camera stream on a grid, 
• Application of CUDA video processing filters, classification and detection inference, 
• Video stabilization processing and video streaming through the network.

Please watch Jetson Xavier Multicamera + AI + Video Stabilization + CUDA Video Processing FIlters demo from RidgeRun :

Demo components:

D3 Engineering-Nvidia-Xavier FPD-Link III interface card

                                   

D3 Engineering-D3RCM-OV10640-953 Rugged Camera Module

The 8 camera streams are downscaled to 480x480 resolution and displayed on a grid. Following are the extra processing is applied to different camera streams:

Camera_1: No extra processing, just normal camera stream. Intended to be used as a point of comparison against the streams with CUDA video processing filters.

Camera_2: Sobel in X-axis CUDA video filter applied with GstCUDA plugin.

Camera_3: Border Enhancement CUDA video filter applied with GstCUDA plugin.

Camera_4: Grayscale CUDA video filter applied with GstCUDA plugin.

Camera_5: No extra processing, just normal camera stream. Intended to be used as a point of comparison against the stream with video stabilization processing.

Camera_6: Video stabilization processing applied with GstNvStabilize plugin. 

Camera_7: InceptionV1 Classification Inference applied with GstInference plugin using GPU accelerated TensorFlow.

Camera_8: TinyYoloV2 Detection Inference applied with GstInference plugin using GPU accelerated TensorFlow.

One individual camera stream selected by the user from the demo menu is streamed to the network using the GstWebRTC plugin and an OpenWebRTC application.

Demo setup, demo features in detail, demo code and performance profiling information are explained in this RidgeRun & D3 Engineering - Nvidia Partner Showcase : Jetson Xavier Multi-Camera AI Demo RidgeRun Developer Wiki.

Contact Us

Please visit our Main Website for the RidgeRun online store or Contact Us for pricing information of the engineering support, product and services.                                                                              

You can also send an email to support@ridgerun.com for a technical support, more information about the features, evaluation version (if available) or for a details about how to sponsor a new feature.

RidgeRun's V4L2 interface for PCIe connected FPGAs : FPGA V4L2 PCIe Driver

HW acceleration is an essential component of modern embedded systems. With ever increasing real-time demands and low power requirements, long are the days where single CPUs systems could fullfill today's market expectations. Among available accelerators, FPGAs typically excel in flexibility and performance, at the cost of integration complexities. It's common to see every FPGA integrated differently in every product, with different interfaces and home-made APIs. Despite this, one common remark that can be observed is that they are typically connected via high bandwidth PCIe.
RidgeRun is developing a single, standard V4L2 interface for PCIe connected FPGAs for a variety of vendors and models. No matter your HW setup, the FPGA is exposed as a combination of camera and display devices. This allows out-of-the-box usage with OpenCV, GStreamer, Libav, browsers and any other standard software that communicates via V4L2 calls.

This V4L2-FPGA driver act as an alternative to solve FPGA-SoC communication in a more standard way, without sacrificing communication performance and make you able to concentrate in the FPGA hardware description. V4L2-FPGA driver allows communicating with an external FPGA using the V4L2 API.

Figure 1. Software stack description using RidgeRun's V4L2 FPGA.

This project consists of three subsystems which allow for the acceleration of algorithms on custom hardware as shown in the following image:

Figure 2: V4L2 Data Flow

Current on-going development is targeted for the PicoEVB Xilinx module on the NVIDIA Xavier. Contact us if you are interested in sponsoring the port to your hardware configuration. 

More details in our developer's wiki : https://developer.ridgerun.com/wiki/index.php?title=V4L2_FPGA

Contact Us

Please visit our main website https://www.ridgerun.com for the RidgeRun online store or https://www.ridgerun.com/contact for pricing information of the engineering support, product and services. You can also send an email to support@ridgerun.com for a technical support, more information about the features, evaluation version (if available) or for a details about how to sponsor a new feature.

GStreamer Video Stabilizer for NVIDIA Jetson Boards

Many applications require the removal of undesired camera movement. Professional video stitching, medical imaging such as colonoscopy or endoscopy and localization of unmanned vehicles are a few examples of use cases that benefit from video stabilization. Unfortunately, this is a very resource consuming technique that may be unfeasible for real time operations on resource constrained systems such as embedded systems.

The following video provides a hands-on overview of GstNvStabilize on the works!

GstNvStabilize is GStreamer based video stabilizer for NVIDIA Jetson boards. It's based on VisionWorks and OpenVX hardware processing units to accelerate the stabilization for real time applications.

Latest v0.4.0 release include:

- Region-of-interest configuration via GStreamer caps - Smoothing level configuration via GStreamer property - Smart compensation limit to avoid black borders - GPU acceleration - Supported platforms: - NVIDIA Jetson Xavier - NVIDIA Jetson TX1/TX2 - NVIDIA Jetson Nano

Learn more in our developer's wiki:
https://developer.ridgerun.com/wiki/index.php?title=GStreamer_Video_Stabilizer_for_NVIDIA_Jetson_
Boards

Purchase directly from our website:
https://shop.ridgerun.com/products/gstnvstabilize?_pos=1&_sid=0951b9cf7&_ss=r 

Contact Us

Please visit our Main Website for the RidgeRun online store or Contact Us for pricing information of 
the engineering support, product and services. 
You can also send an email to support@ridgerun.com for a technical support, more information about 
the features, evaluation version (if available) or for a details about how to sponsor a new feature.

Nvidia Jetson Xavier multi camera Artificial Intelligence demo showcase by RidgeRun

This demo from RidgeRun shows the capabilities of the Jetson Xavier by performing :

• Multi-camera capture through FPD-LINK III with Virtual Channels support, 
• Display of each individual camera stream on a grid, 
• Application of CUDA video processing filters, classification and detection inference, 
• Video stabilization processing and video streaming through the network.

RidgeRun demo screen:

RidgeRun & D3 Engineering Nvidia Partner Showcase Jetson Xavier Multi-Camera AI Demo.

Demo components:

D3 Engineering-Nvidia-Xavier FPD-Link III interface card

                                   

D3 Engineering-D3RCM-OV10640-953 Rugged Camera Module

The 8 camera streams are downscaled to 480x480 resolution and displayed on a grid. Following are the extra processing is applied to different camera streams:

Camera_1: No extra processing, just normal camera stream. Intended to be used as a point of comparison against the streams with CUDA video processing filters.

Camera_2: Sobel in X-axis CUDA video filter applied with GstCUDA plugin.

Camera_3: Border Enhancement CUDA video filter applied with GstCUDA plugin.

Camera_4: Grayscale CUDA video filter applied with GstCUDA plugin.

Camera_5: No extra processing, just normal camera stream. Intended to be used as a point of comparison against the stream with video stabilization processing.

Camera_6: Video stabilization processing applied with GstNvStabilize plugin. 

Camera_7: InceptionV1 Classification Inference applied with GstInference plugin using GPU accelerated TensorFlow.

Camera_8: TinyYoloV2 Detection Inference applied with GstInference plugin using GPU accelerated TensorFlow.

One individual camera stream selected by the user from the demo menu is streamed to the network using the GstWebRTC plugin and an OpenWebRTC application.

Demo setup, demo features in detail, demo code and performance profiling information are explained in this RidgeRun & D3 Engineering - Nvidia Partner Showcase : Jetson Xavier Multi-Camera AI Demo RidgeRun Developer Wiki.

Contact Us

Please visit our Main Website for the RidgeRun online store or Contact Us for pricing information of the engineering support, product and services.                                                                              

You can also send an email to support@ridgerun.com for a technical support, more information about the features, evaluation version (if available) or for a details about how to sponsor a new feature.

GstCUDA: RidgeRun presentation at NVIDIA GTC 2019 on GStreamer and CUDA integration.

RidgeRun engineers presented a GstCUDA, a framework developed by RidgeRun that provides an easy, flexible and powerful integration between GStreamer audio/video streaming infrastructure and CUDA hardware-accelerated video processing at NVIDIA GTC 2019.

GstCUDA: Easy GStreamer and CUDA Integration

Please Watch the Video

For more information please contact us at support@ridgerun.com or for purchase related questions post your inquiry at our Contact Us page.

RidgeRun at GTC 2019 as a NVIDIA Jetson partner

RidgeRun is excited about the new things coming for the NVIDIA Jetson partners and our team will be at  GTC 2019!!

RidgeRun Engineering Manager with Jensen Huang CEO NVIDIA

For more information please contact us at support@ridgerun.com or for purchase related questions post your inquiry at our Contact Us page.

RidgeRun support "Armstrong" Robot at The 2019 FIRST® Robotics competition by providing NVIDIA Jetson support

FIRST Robotics Competition [1] Team 102, The Gearheads, from Somerville High School in NJ, demonstrating the "Armstrong" - The robot for the 2019 season. They hope that the combination of Nvidia TX1, Auvidea J90LC, and drivers from RidgeRun will provide a cost effective, state of the art Computer Vision solution.  Go Gearheads!’

[1] https://www.firstinspires.org/robotics/frc

RidgeRun help getting the kernel built with the IMX219 V4L2 driver for TX1 on the Auvidea J90-LC for state of the art Computer Vision solution for The 2019 FIRST® 2019 season.

Pictures from the FIRST Robotics Competition at Bridgewater, NJ on March 16, 2019.

"Armstrong" - The robot.

Please note the Ridgerun graphic on the "Armstrong" - The robot.

"Armstrong" - The robot in action.

More close-up Ridgerun graphic on the "Armstrong" - The robot

Team 102, The Gearheads, qualified for the finals out of 35, went on to tie in the semi finals and ultimately finished very favorably.

Next competition is the Greater Pittsburgh Regional in PA on Mar 20 - 23.

http://pittsburghfirst.org/2019-greater-pittsburgh-regional/

Watch out! and we will keep updating as the competition progress.

For more information please contact us at support@ridgerun.com or for purchase related questions post your inquiry at our Contact Us page. 

GstInference - Bringing AI to GStreamer

AI is everywhere. Well, maybe this is not entirely true yet, but will be in a few years from now. I heard a prediction that 1,000,000,000 video cameras will be installed by 2020. At RidgeRun we've been working hard to join this revolution, and we want to share some of the technology we're developing with the community.

The truth is that the initial face-off with AI is never easy. A quick Google search reveals a handful of frameworks available to implement a neural network. For the sake of this example suppose you choose TensorFlow. You grasp the whole graph oriented processing paradigm, and finally understand and successfully run the image classification example you are trying out. Cool! This is exactly what your product needs to do. Almost.

Now to fit all the pieces together. Here's a small list of the most immediate tasks for your project:

• Modify the example to receive images from the camera, rather than PNG images.
• Scale down the 1920x1080 camera images to 224x224 needed by the neural network.
• Render a label in the image according to the network prediction.
• Save resulting images in a MP4 file.

If you are familiar with GStreamer, you immediately envisioned a nice, simple pipeline. Something like the following:

Putting it this way doesn't feel that overwhelming after all.

That's exactly where GstInference fits in this whole scenario. It's a set of GStreamer elements that execute AI models on a video stream. We've taken care of all the complexities and boilerplate code so that all the framework details are hidden, and you just link in the inference element as any other. And not only TensorFlow, but Caffe, NCSDK, TensorRT, and others as well (yet to come).

Here's how the whole pipeline looks altogether:

Note how the modular architecture of GStreamer allows you to make the most of the available hardware accelerators in your platform. The scaling could be done using the Image Signal Processor, the encoding using the Video Processing Unit and the inference using the GPU. Let's take it one step further: using the exact same pipeline you can migrate the inference processing from the GPU to the Tensor Processing Unit by simply changing the backend property in GstInference from TensorFlow to TensorRT, for example.

RidgeRun's main design goal is simplicity for the user.

We have made an early release available. Feel free to give it a try and give us your feedback. We'd love to hear from you!

https://developer.ridgerun.com/wiki/index.php?title=GstInference
https://github.com/RidgeRun/gst-inference/

RidgeRun's Sony IMX219 CMOS Image Sensor Linux Driver for NVidia Jetson Xavier and Jetson TX1/TX2

This blog highlights the RidgeRun support for Jetson Xavier and Jetson Tegra platform on developing a CMOS Image Sensor Linux Driver for Sony IMX219.

Driver Features:

• L4T 31.1 and Jetpack 4.1
• V4l2 Media controller driver
• One camera capturing (TODO: to expand to 6 cameras)
• Tested resolution 3280 x 2464 @ 15 fps
• Tested resolution 720p @ 78 fps
• Tested resolution 1640x1232 @ 30 fps
• Tested resolution 820x616 @ 30 fps
• Tested with J20 Auvidea board.
• Capture with v4l2src and also with nvarguscamerasrc using the ISP.

Images attached here are taken during the developing driver for Sony IMX219 image sensor, testing various image capture and display options, performance and latency measurement in our R&D lab located in CostaRica. Various tests are carried out using GStreamer pipelines. 

Enabling and building the driver with Auvidea J20 Expansion board, Example GStreamer pipelines, Performance, Latency measurement details are shared in the RidgeRun developer wiki's mentioned at the end of this blog.

RidgeRun's Sony IMX219 Linux driver for Jetson Xavier

Output image of the IMX219 camera sensor image capture @1640x1232 resolution with nvcamerasrc GStreamer element for the Jetson Xavier platform:

RidgeRun's Sony IMX219 Linux driver latency measurements on Jetson Xavier

Frames showing how the glass to glass latency measurement method is setup by our team while testing for image capture using IMX219 camera mode at 3280x2464@16fps resolution for Xavier platform.

 

Glass to glass latency measured is 215 ms (07:772 minus 07:557).Time readings can be seen in the displays.

RidgeRun's Sony IMX219 Linux driver for Jetson TX1

Image below is showing the IMX219 capture @1640x1232 resolution with nvcamerasrc on the Jetson TX1 platform. Camera aimed at computer monitor on the left which is reflecting the wall and ceiling shown on the right:

RidgeRun's Sony IMX219 Linux driver latency measurements on Jetson TX1

Image below is captured while measuring the Jetson TX1 glass to glass latency for 1080p 30fps IMX219 camera mode:

 

Glass to glass latency measured is 130 ms ((13.586 minus 13.456).Time readings can be seen in the displays.

You can find more information about the driver in these developer wiki's from RidgeRun : 

Sony IMX219 Linux Driver for Jetson Xavier

Sony IMX219 Linux driver for Jetson TX1

For technical information, please email to us at support@ridgerun.com or for purchase related questions post your inquiry at our Contact Us page.

RidgeRun's USB Video Class Gadget Library - LibGUVC v1.4.0 - NVidia Xavier and NXP iMX Support

The UVC Video Class Gadget Library or libguvc for short is a platform agnostic library that simplifies the development of UVC based gadget devices by encapsulating the most of the UVC communication leaving just the basic setup to the user. The USB video class gadget runs on top of the UVC function driver in the user space and takes care of the communication between the user application and the linux driver stack.

Release of libGUVC v1.4.0, is now supporting NVIDIA Jetson platform along with previously supported NXP-iMX6 family of processors, thanks to the new bulk transfer support.

It has never been easier to implement a UVC application on your hardware, libGUVC make it easy to interact with the UVC driver and expose a variety of useful features such as:

•     USB 2.0 and USB 3.0 support
•     Isochronous and bulk endpoint support
•     YUV2, MJPEG and H264 video streaming support
•     Extension Unit support
•     MMAP and UserPtr support.

Since the driver is agnostic to the platform, you can run it on almost any platform with quality USB and UVC drivers, making it really simple to convert it into a UVC capable device.

 libGuvc in action - Running libGuvc on NVidia Xavier

 

 libGuvc in action - Running libGuvc on NXP-iMX6

 

You can find more information about the library in this developer wiki from RidgeRun Engineering : USB Video Class Gadget Library - libguvc

For more technical information please contact us at support@ridgerun.com. Please post your inquiry at our Contact Us page for purchase related questions and also since the libguvc is platform agnostic, you can request for a custom demo image for any other platform using this link.

RidgeRun supporting The 2019 FIRST® Robotics by providing NVIDIA Jetson support

RidgeRun love helping academic projects! RidgeRun is proud to help the teams on FIRST Robotics Competition [1] with software for embedded systems to improve the acquisition, processing and analysis of Audio and Video signals!

[1] https://www.firstinspires.org/robotics/frc

Michael, Amanda, and Emily (in the pic below) of FIRST Robotics Competition Team 102, The Gearheads, from Somerville High School in NJ, investigate the NVIDIA Jetson technology from RidgeRun.  They hope that the combination of Nvidia TX1, Auvidea J90LC, and drivers from RidgeRun will provide a cost effective, state of the art Computer Vision solution for the 2019 season. Go Gearheads!’

RidgeRun help getting the kernel built with the IMX219 V4L2 driver for TX1 on the Auvidea J90-LC for state of the art Computer Vision solution for The 2019 FIRST® 2019 season.

Watch out! and we will keep updating as the Competition progress.

For more information please contact us at support@ridgerun.com or for purchase related questions post your inquiry at our Contact Us page.

RidgeRun - NVIDIA Xavier - Deep Learning Tutorials using Jetson Inference

Jetson-inference is a training guide for inference on the TX1 and TX2 using NVIDIA Deep Learning GPU Training System (DIGITS). 

This blog details the summary of original Jetson-inference training from NVIDIA with a focus on inference part.

You can learn about following details in this developer wiki from RIdgeRun Engineering : NVIDIA Xavier - Deep Learning - Deep Learning Tutorials - Jetson Inference

Building jetson-inference.

Classifying Images with ImageNet.

Locating Object Coordinates using DetectNet.

Image Segmentation with SegNet.

and run a Live Demo.

With jetson-inference you can deploy deep learning examples on the NVIDIA Xavier in a matter of minutes. An example application is shown below.

The input is an image and it outputs the most likely class and the probability that the image belongs to that class using ImageNet classification network. ImageNet is a classification network trained with a database of 1000 objects.

Fig1: imagenet-console output image

It detects the image as 'Boston bull, Boston terrier' with imagenet class id of 0195 at 96.305% classification accuracy. Image recognition networks output a class probabilities corresponding to the entire input image.

Detection networks, on the other hand, find where in the image those objects are located. DetectNet accepts an input image, and outputs the class and coordinates of the detected bounding boxes.

                            Fig2: detectnet-console output image using coco-dog pretrained model


If you are new to the Xavier or planning on getting one, please visit our Jetson Xavier Wiki page.

Article related : 

Read this blog on deep reinforcement learning Deep Reinforcement Learning on the Jetson Xavier

For more information please contact us at support@ridgerun.com