{"id":33374,"date":"2020-06-04T22:47:41","date_gmt":"2020-06-04T14:47:41","guid":{"rendered":"\/blog\/?p=33374"},"modified":"2022-06-02T11:57:34","modified_gmt":"2022-06-02T03:57:34","slug":"nvidia-jetson-nano-and-jetson-xavier-nx-comparison-specifications-benchmarking-container-demos-and-custom-model-inference","status":"publish","type":"post","link":"https:\/\/www.seeedstudio.com\/blog\/2020\/06\/04\/nvidia-jetson-nano-and-jetson-xavier-nx-comparison-specifications-benchmarking-container-demos-and-custom-model-inference\/","title":{"rendered":"NVIDIA Jetson Nano and Jetson Xavier NX Comparison: Specifications, Benchmarking, Container Demos, and Custom Model Inference"},"content":{"rendered":"\n<figure class=\"wp-block-image\"><a href=\"https:\/\/www.seeedstudio.com\/tag\/nvidia.html\"><img decoding=\"async\" src=\"https:\/\/media-cdn.seeedstudio.com\/media\/wysiwyg\/nvidia\/_banner.jpg\" alt=\"\"\/><\/a><figcaption><a href=\"https:\/\/www.seeedstudio.com\/tag\/nvidia.html\">Latest new arrivals for Edge AI powered by NVIDIA Embedded System<br><\/a><\/figcaption><\/figure>\n\n\n\n<h2 class=\"wp-block-heading\">New release for NVIDIA Jetson Nano and Jetson Xavier NX<\/h2>\n\n\n\n<p>reComputer Jetson series are compact edge computers built with NVIDIA advanced AI embedded systems: <\/p>\n\n\n\n<ul class=\"wp-block-list\"><li><strong><a href=\"https:\/\/www.seeedstudio.com\/Jetson-10-1-H0-p-5335.html\">reComputer J1010, J1020 (built with Jetson Nano Module 4GB)<\/a> <\/strong><\/li><li><strong><a href=\"https:\/\/www.seeedstudio.com\/Jetson-20-1-H1-p-5328.html\">reComputer J2011, J2012 (built with Jetson Xavier NX Module 16GB\/8GB)<\/a>.<\/strong> <\/li><\/ul>\n\n\n\n<p>With rich extension modules, industrial peripherals, thermal management, reComputer is ready to help you accelerate and scale the next-gen AI product by deploying popular DNN models and ML frameworks to the edge and inferencing with high performance, for tasks like real-time classification and object detection, pose estimation, semantic segmentation, and natural language processing (NLP).<\/p>\n\n\n\n<p>With Jetson Nano, developers can use highly accurate pre-trained models from TAO Toolkit and deploy with DeepStream. Jetson Nano can achieve 11 FPS for PeopleNet- ResNet34&nbsp; of People Detection, 19 FPS for<a href=\"https:\/\/wiki.seeedstudio.com\/DashCamNet-with-Jetson-Xavier-NX-Multicamera\/\">&nbsp;DashCamNet-ResNet18 of Vehicle Detection<\/a>, and 101 FPS for FaceDetect-IR-ResNet18 of Face Detection. Benchmark details can be found on&nbsp;<a href=\"https:\/\/developer.nvidia.com\/deepstream-sdk\">NVIDIA\u00ae\u2019s DeepStream SDK website<\/a>.<\/p>\n\n\n\n<p>With Jetson Xavier NX, developers can use highly accurate pre-trained models from TAO Toolkit and deploy with DeepStream. Jetson Xavier NX can achieve 172 FPS for PeopleNet- ResNet34&nbsp; of People Detection, 274 FPS for<a href=\"https:\/\/wiki.seeedstudio.com\/DashCamNet-with-Jetson-Xavier-NX-Multicamera\/\">&nbsp;DashCamNet-ResNet18 of Vehicle Detection<\/a>, and 1126 FPS for FaceDetect-IR-ResNet18 of Face Detection. Benchmark details can be found on&nbsp;<a href=\"https:\/\/developer.nvidia.com\/deepstream-sdk\">NVIDIA\u00ae\u2019s DeepStream SDK website<\/a>.<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img fetchpriority=\"high\" decoding=\"async\" width=\"1030\" height=\"315\" src=\"https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2022\/04\/image-28-1030x315.png\" alt=\"\" class=\"wp-image-63636\" srcset=\"https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2022\/04\/image-28-1030x315.png 1030w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2022\/04\/image-28-300x92.png 300w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2022\/04\/image-28-768x235.png 768w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2022\/04\/image-28-1536x469.png 1536w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2022\/04\/image-28-2048x626.png 2048w, https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2022\/04\/image-28-1024x313.png 1024w\" sizes=\"(max-width: 1030px) 100vw, 1030px\" \/><figcaption>Jetson series benchmark for Deepstream: <a href=\"https:\/\/developer.nvidia.com\/deepstream-sdk\">end-to-end application performance from data ingestion, decoding, image processing to inference<\/a><\/figcaption><\/figure>\n\n\n\n<p>The following article is a transcript from the video published on <a href=\"https:\/\/www.youtube.com\/watch?v=s0lW3Hkcx-k&amp;feature=emb_title\">Hardware.ai  YouTube channel<\/a>. It also appeared on <a href=\"https:\/\/www.hackster.io\/dmitrywat\/jetson-xavier-nx-vs-jetson-nano-detailed-comparison-aa9cd7\">Hackster.io<\/a><\/p>\n\n\n\n<p>Don&#8217;t miss out on&nbsp;<a href=\"https:\/\/www.seeedstudio.com\/NVIDIA-Jetson-Xavier-NX-Developer-Kit-p-4573.html\">NVIDIA Jetson Xavier NX<\/a> and&nbsp;<a href=\"https:\/\/www.seeedstudio.com\/NVIDIA-Jetson-Nano-Development-Kit-B01-p-4437.html\">NVIDIA Jetson Nano<\/a>&nbsp;at Seeed! <\/p>\n\n\n\n<figure class=\"wp-block-embed is-type-video is-provider-youtube wp-block-embed-youtube wp-embed-aspect-16-9 wp-has-aspect-ratio\"><div class=\"wp-block-embed__wrapper\">\n<iframe title=\"Jetson Xavier NX versus Jetson Nano - Benchmarks and Custom Model Inference\" width=\"640\" height=\"360\" src=\"https:\/\/www.youtube.com\/embed\/s0lW3Hkcx-k?feature=oembed\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share\" referrerpolicy=\"strict-origin-when-cross-origin\" allowfullscreen><\/iframe>\n<\/div><\/figure>\n\n\n\n<p>Today we\u2019re going to have a good look at a new development kit from NVIDIA &#8211;&nbsp;<strong>Xavier NX<\/strong>&nbsp;and compare it to another dev kit from NVIDIA,&nbsp;<strong>Jetson Nano<\/strong>. The compute module Xavier NX was announced on November 6, 2019, but the development kit, which includes the module and reference carrier board was announced just half a year later, on May 14, 2020. This is the opposite situation of what happened with Nvidia Jetson Nano &#8211; in that case, development kit came first and then the compute module became available for purchase.<\/p>\n\n\n\n<p>So, how does Xavier NX compare to Jetson Nano? The price difference is significant,&nbsp;<strong>99 USD vs. 399 USD<\/strong>&nbsp;or about 400 percent. Is there a clear threshold for when you know that Jetson Nano is not enough for your application and you need to up your game a level?<\/p>\n\n\n\n<div class=\"wp-block-image\"><figure class=\"aligncenter is-resized\"><img decoding=\"async\" src=\"https:\/\/hackster.imgix.net\/uploads\/attachments\/1134544\/6938280971_326960b48a_z_dlbTCS80Ft.jpg?auto=compress%2Cformat&amp;w=740&amp;h=555&amp;fit=max\" alt=\"\" width=\"214\" height=\"300\"\/><\/figure><\/div>\n\n\n\n<p>Before we start our comparison, we should note that these two products being targeted for different target groups &#8211; Nvidia Jetson Nano is for&nbsp;<strong>makers and STEM education<\/strong>, while Xavier NX is more geared towards&nbsp;<strong>professional and commercial use<\/strong>. It was very obvious from the sample applications that Nvidia released on product\u2019s launch &#8211; for Jetson Nano it was Jetboat with a series of user-friendly notebooks and for Jetson Xavier NX it was a demo of cloud-native applications, which appeals to commercial users. That still warrants a comparison to be made &#8211; think about it as Xavier NX being a sports car with two turbo engines and Jetson Nano is a more down to earth sedan. We don\u2019t always go for the fastest car available, even if we have the money, there are many other considerations to take.<\/p>\n\n\n\n<p>In this article, we\u2019ll go over&nbsp;<\/p>\n\n\n\n<ul class=\"wp-block-list\"><li><strong>Hardware Overview <\/strong><\/li><li><strong>Specs <\/strong><\/li><li><strong>Benchmarking<\/strong><\/li><li><strong>Cloud-Native container demo&nbsp;<\/strong><\/li><\/ul>\n\n\n\n<p><strong>Custom Model Inference<\/strong>&nbsp;<\/p>\n\n\n\n<p>Let&#8217;s look into these small but quite important details sometimes overlooked in other reviews.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"toc-hardware-0\">Hardware and Specification Comparison of NVIDIA Jetson Nano and Jetson Xavier NX<\/h3>\n\n\n\n<p>Let\u2019s start by looking at two dev kits side by side. I\u2019ll skip the unpacking, because, well, you know how to open a box without me. If not, please leave a comment below, I\u2019ll make another tutorial soon. Fist, the similarities. Both dev boards are similar in size and the modules are exactly the same size and form factor. Which is great for developers, it means that during developments stage swapping Nano module for Xavier NX won\u2019t require changing connectors and physical design of carrier board (providing they use Rev B01 of Jetson Nano Dev Kit).<\/p>\n\n\n\n<figure class=\"wp-block-image\"><img decoding=\"async\" src=\"https:\/\/hackster.imgix.net\/uploads\/attachments\/1134547\/jetson_nx_briefing_16_vmxipsyqVB.png?auto=compress%2Cformat&amp;w=740&amp;h=555&amp;fit=max\" alt=\"\"\/><\/figure>\n\n\n\n<p>Both carrier boards have Gigabit Ethernet Jack, 4 USB 3 ports (although for Xavier NX it is USB 3.1 and for Jetson Nano it is 3.0), HDMI port and display ports.<\/p>\n\n\n\n<p>Now the differences. Xavier NX dev kit board has&nbsp;<strong>one M2 Key E and one M2 Key M connector<\/strong>, with M2 Key E slot already occupied with BT\/WiFi module. M2 Key M can be used for attaching NVME SSD. Jetson Nano carrier board only has&nbsp;<strong>one M2 Key E connector<\/strong>. The Jetson Nano carrier board I have here is A02 carrier board, the one that originally went on sale when Jetson Nano dev kit was just released &#8211; it has some differences from B01 Carrier Board, notably B01 has two CSI-2 camera interface, same as Xavier NX carrier board, while A02 you see here just has one.<\/p>\n\n\n\n<figure class=\"wp-block-image\"><img decoding=\"async\" src=\"https:\/\/hackster.imgix.net\/uploads\/attachments\/1134548\/capture1_GzYBNunhYY.PNG?auto=compress%2Cformat&amp;w=740&amp;h=555&amp;fit=max\" alt=\"\"\/><\/figure>\n\n\n\n<p>Xavier NX has an active cooling installed, while Jetson Nano only has a heatsink. It\u2019s because Xavier is much more power-hungry than it\u2019s a younger cousin &#8211; the dev kit cannot be powered by 5V USB and&nbsp;<strong>requires 19V power supply<\/strong>, which fortunately is included in the box. The reference carrier board of Xavier NX has more well-thought design with a little plastic base, that serves two purposes: protects the circuits from directly touching the surface of your workplace and holds two antennas. A slight detail, but very nice.<\/p>\n\n\n\n<p>You can also check our blog <a href=\"https:\/\/www.seeedstudio.com\/blog\/2020\/01\/16\/new-revision-of-jetson-nano-dev-kit-now-supports-new-jetson-nano-module\/\">New Revision of Jetson Nano Dev Kit \u2013 Now supports the new Jetson Nano Module<\/a> to have a detailed look at how the new B01 revised carrier board id different from the A02 board. <\/p>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"toc-specs-1\">Specs<\/h3>\n\n\n\n<p>Let\u2019s quickly compare the specs now, using the data from&nbsp;<a href=\"https:\/\/developer.nvidia.com\/embedded\/develop\/hardware\">official NVIDIA website<\/a>.<\/p>\n\n\n\n<p>Apart from very obvious things, such as Xavier NX having CPU with more cores and higher clock rate and more RAM(faster too, 25.6GB\/s vs 51.2GB\/s), it is the AI Performance column that worth paying more attention too. Before you go, \u201cHoly cow, the AI performance of Xavier NX is 44 times higher, than of Jetson Nano\u201d(21\/0.472~44) it is worth noting that the <strong>comparison is between different units TOPS vs TFLOPS.<\/strong><\/p>\n\n\n\n<figure class=\"wp-block-image\"><img decoding=\"async\" src=\"https:\/\/hackster.imgix.net\/uploads\/attachments\/1134549\/untitled_project_(time_0_06_2000)_srvfMlxeHL.jpg?auto=compress%2Cformat&amp;w=740&amp;h=555&amp;fit=max\" alt=\"\"\/><\/figure>\n\n\n\n<p>The reason for that is for Xavier NX, the compute of&nbsp;<strong>NVDLA Engines<\/strong>&nbsp;is included in that very impressive number. Jetson Nano (and older members of Jetson family, TX1\/TX2) only have GPU for accelerating ML inference, which is optimized for floating-point operations. NVDLA Engines on the other hand are different beasts entirely and are ASICs or Application Specific Integrated Circuits, more akin to Google TPU\/Intel Movidius chips.<\/p>\n\n\n\n<figure class=\"wp-block-image\"><img decoding=\"async\" src=\"https:\/\/hackster.imgix.net\/uploads\/attachments\/1134588\/untitled_project_(time_0_07_0303)_ID1TWadLOc.jpg?auto=compress%2Cformat&amp;w=740&amp;h=555&amp;fit=max\" alt=\"\"\/><\/figure>\n\n\n\n<p>They&nbsp;<strong>excel at running CNN inference in INT8 precision<\/strong>, doing that task faster and more energy efficient than GPUs &#8211; the downside being that they are not as the general purpose when it comes to different network architecture support. NVIDIA realized that GPUs alone cannot beat highly specialized hardware and decided to take the best of both worlds by having both advanced&nbsp;<strong>384-core NVIDIA Volta\u2122 GPU<\/strong>&nbsp;with 48 Tensor Cores and dedicated CNN accelerators in the new module.<\/p>\n\n\n\n<figure class=\"wp-block-image\"><img decoding=\"async\" src=\"https:\/\/hackster.imgix.net\/uploads\/attachments\/1134589\/untitled_project_(time_0_07_2117)_D6lMaiHXq3.jpg?auto=compress%2Cformat&amp;w=740&amp;h=555&amp;fit=max\" alt=\"\"\/><\/figure>\n\n\n\n<p>Speaking of Tensor Cores, we also see that these are missing from Nvidia Jetson Nano\u2019s GPU. Now, what on Earth is Tensor core? That\u2019s exactly the title of an article on the first page of Google Search results for tensor core. Comparing to CUDA cores, CUDA cores operate on a per-calculation basis, each individual CUDA core can perform one precise calculation per revolution of the GPU. As a result, clock speed plays a major role in the performance of CUDA, as well as the mass of CUDA cores available on the card. Tensor cores, on the other hand, can calculate with an entire 4&#215;4 matrice operation being calculated per clock.<\/p>\n\n\n\n<p>Look at the animation here to get a sort of intuitive understanding of what is going on in regular CUDA core and Tensor core.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"toc-benchmarking-2\">Benchmarking<\/h3>\n\n\n\n<p>Alright, that was a brisk, but invigorating walk through forests of high-performance computing. If you\u2019re an engineer like me, you want to know how exactly all of that translates to inference performance. This time NVIDIA created&nbsp;<a href=\"https:\/\/github.com\/NVIDIA-AI-IOT\/jetson_benchmarks\">a dedicated repository<\/a>&nbsp;with easily downloadable and executable benchmarks, a decision I can applaud to. So, I ran the benchmarks from this repository on both Nvidia Jetson Nano and Xavier NX. Quite unsurprisingly the results were very close to these from Nvidia Blog article.<\/p>\n\n\n\n<figure class=\"wp-block-image\"><img decoding=\"async\" src=\"https:\/\/hackster.imgix.net\/uploads\/attachments\/1134554\/capture6_uI5pZjMST2.PNG?auto=compress%2Cformat&amp;w=740&amp;h=555&amp;fit=max\" alt=\"\"\/><\/figure>\n\n\n\n<p>Something worth paying attention to is that for Xavier NX the total FPS is the sum of FPS obtained from running model on two DLA and GPU simultaneously. You can find that by looking at the content of benchmark.py.<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code># Note: GPU, DLA latencies are measured in miliseconds, FPS = Frames per Secondprint(benchmark_table&#91;&#91;'Model Name', 'DLA0 (ms)', 'DLA1 (ms)', 'FPS']])<\/code><\/pre>\n\n\n\n<p>You can modify it to print out the total FPS and inference time for each device (for Xavier NX), like that.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"toc-container-demo-3\">Container demo<\/h3>\n\n\n\n<p>The second test will be using much touted new features in Jetpack 4.4, \u201ccloud-native\u201d. Basically, it is about bringing containerization, and orchestration from servers to edge devices. Why do that?&nbsp;To simplify the continued development. Containers are self-contained packages, that include all the necessary environment to run an application. The main selling point of containerization is that they make upgrades easier &#8211; because they\u2019re self-contained, you don\u2019t need to worry about updating one package will cause versioning bugs in your whole system. Nvidia prepared a \u201ccustomer service\u201d robot demo to showcase the new container management system and hardware capabilities of Xavier NX. Unfortunately, because 3 of 4 containers for that demo have TensorRT engine files built for Jetson AGX Xavier and Jetson Xavier NX and can be run on Jetson AGX Xavier or Jetson Xavier NX only. So, we\u2019ll download and try another container,&nbsp;Deepstream-L4T, and run sample application from within the container. Sample application are already precompiled and ready to run within containers.  Unfortunately, they hard-coded to look for their respective config file in the folder where you run the application, which might cause some confusion.<\/p>\n\n\n\n<p>Download the container with<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>sudo docker pull nvcr.io\/nvidia\/deepstream-l4t:5.0-dp-20.04-samples<\/code><\/pre>\n\n\n\n<p>Allow external applications to connect to the host&#8217;s X display<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>xhost +<\/code><\/pre>\n\n\n\n<p>After that run the container with<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>sudo docker run -it --rm --net=host --runtime nvidia  -e DISPLAY=$DISPLAY -w \/opt\/nvidia\/deepstream\/deepstream-5.0 -v \/tmp\/.X11-unix\/:\/tmp\/.X11-unix nvcr.io\/nvidia\/deepstream-l4t:5.0-dp-20.04-samples<\/code><\/pre>\n\n\n\n<p>Then move to the folder with app config file<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>cd sources\/apps\/sample_apps\/deepstream-test3<\/code><\/pre>\n\n\n\n<p>And execute the test app &#8211; copy and paste the file name to run on multiple streams:<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>deepstream-test3-app file:\/\/\/opt\/nvidia\/deepstream\/deepstream-5.0\/samples\/streams\/sample_720p.mp4<\/code><\/pre>\n\n\n\n<p>we will go for deepstream-apptest3 and run it with 720p video. Jetson Nano can only run smoothly one video stream, once we add the second one, performance drops significantly. For Jetson Xavier we see a similar picture, once we add the second stream, it starts dropping frames.<\/p>\n\n\n\n<figure class=\"wp-block-image\"><img decoding=\"async\" src=\"https:\/\/hackster.imgix.net\/uploads\/attachments\/1134586\/untitled_project_(time_0_13_0312)_eWxGqkd1su.jpg?auto=compress%2Cformat&amp;w=740&amp;h=555&amp;fit=max\" alt=\"\"\/><\/figure>\n\n\n\n<p>Upon checking with jetson_stats we see that CPU and GPU load never reaches 100 percent, so possibly the performance bottleneck here is in SD card reading speed. For their &#8220;service robot&#8221; demo application that runs 4 containers simultaneously it is necessary to use NVME SSD.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"toc-custom-model-inference-4\">Custom model inference<\/h3>\n\n\n\n<p>As for our final comparison, let\u2019s try stepping aside from the demos Nvidia have provided and using&nbsp;<strong>a model we trained ourselves<\/strong>. In the end, this is the real test of performance and user-friendliness &#8211; the results you can achieve, as opposed to carefully optimized demos. We will use&nbsp;<a href=\"https:\/\/github.com\/AIWintermuteAI\/aXeleRate\">aXeleRate, a Keras-based framework for AI on the edge<\/a>. Its purpose is to simplify the training and conversion of models to be run on various edge devices, such as K210, Edge TPU, Android\/Raspberry Pi, and Nvidia dev boards. NVIDIA\u2019s model optimization toolkit,&nbsp;TensorRT&nbsp;is very different from other toolkits for model conversion, such as nncase, Google Coral converter. Unlike the rest of them, with TensorRT you need to optimize model on the target device, since optimizations depend on the target device\u2019s architecture. TensorRT definitely deserves a video &#8211; possibly a video series of its own, and I will make one in near future.<\/p>\n\n\n\n<figure class=\"wp-block-image\"><img decoding=\"async\" src=\"https:\/\/hackster.imgix.net\/uploads\/attachments\/1134561\/1_7zvheeyhihbqjo_pzlwkww_G3X95WDsIx.png?auto=compress%2Cformat&amp;w=740&amp;h=555&amp;fit=max\" alt=\"\"\/><\/figure>\n\n\n\n<p>We will use&nbsp;<a href=\"https:\/\/colab.research.google.com\/github\/AIWintermuteAI\/aXeleRate\/blob\/master\/resources\/aXeleRate_standford_dog_classifier.ipynb\">NASNetMobile trained on Stanford dog breeds dataset<\/a>&nbsp;&#8211; go through the steps in the notebook, remember to modify the config file to enable the conversion to.onnx format!<\/p>\n\n\n\n<figure class=\"wp-block-image\"><img decoding=\"async\" src=\"https:\/\/hackster.imgix.net\/uploads\/attachments\/1134590\/capture7_WYW0rd9XSv.PNG?auto=compress%2Cformat&amp;w=740&amp;h=555&amp;fit=max\" alt=\"\"\/><\/figure>\n\n\n\n<p>After training is done download the.onnx file to Jetson and run onnx_to_trt.py file in the example_scripts\/nvidia_jetson\/classifier folder.<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>onnx_to_trt.py -- onnx (path to .onnx file) --precision FP32<\/code><\/pre>\n\n\n\n<p>The script will convert.onnx model to serialized.plan file which has a model graph optimized for target device &#8211; meaning that you&nbsp;<strong>will not<\/strong>&nbsp;be able to run.plan file created on Xavier NX on Jetson Nano and vice-versa.<\/p>\n\n\n\n<p>Now we can run classifier_video.py on a sample video file.<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>classifier_video.py --model (path to .plan file) --labels (path to .txt file with labels) --source (path to video file)<\/code><\/pre>\n\n\n\n<p>You can use<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>classifier_video.py --model (path to .plan file) --labels (path to .txt file with labels) --source 0<\/code><\/pre>\n\n\n\n<p>to run real-time inference from USB web camera.<\/p>\n\n\n\n<p>Jetson Nano averages FPS around 15 frames per second and Jetson Xavier can process the video with adorable dogs at ~30 frames per second.<\/p>\n\n\n\n<figure class=\"wp-block-image\"><img decoding=\"async\" src=\"https:\/\/hackster.imgix.net\/uploads\/attachments\/1134591\/untitled_project_(time_0_15_0902)_Zf3vJlllYI.jpg?auto=compress%2Cformat&amp;w=740&amp;h=555&amp;fit=max\" alt=\"\"\/><\/figure>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"toc-conclusion-5\">Conclusion<\/h3>\n\n\n\n<p>In conclusion, <strong>new Nvidia Jetson Xavier NX is a beast<\/strong>. <\/p>\n\n\n\n<p>It\u2019s&nbsp;power-hungry, but if performance is what you\u2019re aiming for then it is the best module you can get at this footprint and price. If you are developing an application that requires&nbsp;<strong>processioning multiple video streams<\/strong>&nbsp;at high resolution while performing&nbsp;<strong>ASR\/NLP or another GPU related tasks<\/strong>&nbsp;(CUDA enabled SLAM for example), then it\u2019s deep learning accelerators can take on CNN inference and leave GPU for other tasks &#8211; something that older TX2 or Jetson Nano are not capable of. For makers and hobbyists &#8211; well, if you\u2019re buying it, make sure you already thought of&nbsp;an application that requires that much horsepower&nbsp;under the trunk, and also you have enough&nbsp;<strong>technical knowledge<\/strong>&nbsp;to eliminate any performance bottlenecks if these are present.<\/p>\n\n\n\n<p><\/p>\n","protected":false},"excerpt":{"rendered":"<p>New release for NVIDIA Jetson Nano and Jetson Xavier NX reComputer Jetson series are compact<\/p>\n","protected":false},"author":200,"featured_media":33375,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_lmt_disableupdate":"","_lmt_disable":"","_price":"","_stock":"","_tribe_ticket_header":"","_tribe_default_ticket_provider":"","_tribe_ticket_capacity":"0","_ticket_start_date":"","_ticket_end_date":"","_tribe_ticket_show_description":"","_tribe_ticket_show_not_going":false,"_tribe_ticket_use_global_stock":"","_tribe_ticket_global_stock_level":"","_global_stock_mode":"","_global_stock_cap":"","_tribe_rsvp_for_event":"","_tribe_ticket_going_count":"","_tribe_ticket_not_going_count":"","_tribe_tickets_list":"[]","_tribe_ticket_has_attendee_info_fields":false,"iawp_total_views":0,"footnotes":""},"categories":[1],"tags":[3164,1312,1824,1313,3198],"class_list":["post-33374","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-news","tag-benchmark","tag-nvidia","tag-nvidia-jetson","tag-nvidia-jetson-nano","tag-nvidia-jetson-xavier-nx"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v24.0 - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>NVIDIA Jetson Nano and Jetson Xavier NX Comparison: Specifications, Benchmarking, Container Demos, and Custom Model Inference - Latest News from Seeed Studio<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.seeedstudio.com\/blog\/2020\/06\/04\/nvidia-jetson-nano-and-jetson-xavier-nx-comparison-specifications-benchmarking-container-demos-and-custom-model-inference\/\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"NVIDIA Jetson Nano and Jetson Xavier NX Comparison: Specifications, Benchmarking, Container Demos, and Custom Model Inference - Latest News from Seeed Studio\" \/>\n<meta property=\"og:description\" content=\"New release for NVIDIA Jetson Nano and Jetson Xavier NX reComputer Jetson series are compact\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.seeedstudio.com\/blog\/2020\/06\/04\/nvidia-jetson-nano-and-jetson-xavier-nx-comparison-specifications-benchmarking-container-demos-and-custom-model-inference\/\" \/>\n<meta property=\"og:site_name\" content=\"Latest News from Seeed Studio\" \/>\n<meta property=\"article:published_time\" content=\"2020-06-04T14:47:41+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2022-06-02T03:57:34+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2020\/06\/7421591260005_.pic_hd.jpg\" \/>\n\t<meta property=\"og:image:width\" content=\"1280\" \/>\n\t<meta property=\"og:image:height\" content=\"720\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/jpeg\" \/>\n<meta name=\"author\" content=\"Elaine Wu\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"Elaine Wu\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"13 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/www.seeedstudio.com\/blog\/2020\/06\/04\/nvidia-jetson-nano-and-jetson-xavier-nx-comparison-specifications-benchmarking-container-demos-and-custom-model-inference\/\",\"url\":\"https:\/\/www.seeedstudio.com\/blog\/2020\/06\/04\/nvidia-jetson-nano-and-jetson-xavier-nx-comparison-specifications-benchmarking-container-demos-and-custom-model-inference\/\",\"name\":\"NVIDIA Jetson Nano and Jetson Xavier NX Comparison: Specifications, Benchmarking, Container Demos, and Custom Model Inference - Latest News from Seeed Studio\",\"isPartOf\":{\"@id\":\"https:\/\/www.seeedstudio.com\/blog\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\/\/www.seeedstudio.com\/blog\/2020\/06\/04\/nvidia-jetson-nano-and-jetson-xavier-nx-comparison-specifications-benchmarking-container-demos-and-custom-model-inference\/#primaryimage\"},\"image\":{\"@id\":\"https:\/\/www.seeedstudio.com\/blog\/2020\/06\/04\/nvidia-jetson-nano-and-jetson-xavier-nx-comparison-specifications-benchmarking-container-demos-and-custom-model-inference\/#primaryimage\"},\"thumbnailUrl\":\"https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2020\/06\/7421591260005_.pic_hd.jpg\",\"datePublished\":\"2020-06-04T14:47:41+00:00\",\"dateModified\":\"2022-06-02T03:57:34+00:00\",\"author\":{\"@id\":\"https:\/\/www.seeedstudio.com\/blog\/#\/schema\/person\/61c04bed5bbe2d098f04195c6e48fb11\"},\"breadcrumb\":{\"@id\":\"https:\/\/www.seeedstudio.com\/blog\/2020\/06\/04\/nvidia-jetson-nano-and-jetson-xavier-nx-comparison-specifications-benchmarking-container-demos-and-custom-model-inference\/#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/www.seeedstudio.com\/blog\/2020\/06\/04\/nvidia-jetson-nano-and-jetson-xavier-nx-comparison-specifications-benchmarking-container-demos-and-custom-model-inference\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/www.seeedstudio.com\/blog\/2020\/06\/04\/nvidia-jetson-nano-and-jetson-xavier-nx-comparison-specifications-benchmarking-container-demos-and-custom-model-inference\/#primaryimage\",\"url\":\"https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2020\/06\/7421591260005_.pic_hd.jpg\",\"contentUrl\":\"https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2020\/06\/7421591260005_.pic_hd.jpg\",\"width\":1280,\"height\":720,\"caption\":\"NVIDIA Jetson Nano and Jetson Xavier NX Comparison: Specifications, Benchmarking, Container Demos, and Custom Model Inference\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/www.seeedstudio.com\/blog\/2020\/06\/04\/nvidia-jetson-nano-and-jetson-xavier-nx-comparison-specifications-benchmarking-container-demos-and-custom-model-inference\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\/\/www.seeedstudio.com\/blog\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"NVIDIA Jetson Nano and Jetson Xavier NX Comparison: Specifications, Benchmarking, Container Demos, and Custom Model Inference\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/www.seeedstudio.com\/blog\/#website\",\"url\":\"https:\/\/www.seeedstudio.com\/blog\/\",\"name\":\"Latest News from Seeed Studio\",\"description\":\"Emerging IoT, AI and Autonomous Applications on the Edge\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/www.seeedstudio.com\/blog\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/www.seeedstudio.com\/blog\/#\/schema\/person\/61c04bed5bbe2d098f04195c6e48fb11\",\"name\":\"Elaine Wu\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\/\/www.seeedstudio.com\/blog\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/secure.gravatar.com\/avatar\/184af8ef71f0d6b64c276f9bb38b992e?s=96&r=g\",\"contentUrl\":\"https:\/\/secure.gravatar.com\/avatar\/184af8ef71f0d6b64c276f9bb38b992e?s=96&r=g\",\"caption\":\"Elaine Wu\"},\"description\":\"Head of AI Robotics @seeed Every day holds new magic \u2728 on ne sait jamais\u2601\ufe0f\",\"sameAs\":[\"https:\/\/www.linkedin.com\/in\/elaine1994\/\"],\"url\":\"https:\/\/www.seeedstudio.com\/blog\/author\/elaine\/\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"NVIDIA Jetson Nano and Jetson Xavier NX Comparison: Specifications, Benchmarking, Container Demos, and Custom Model Inference - Latest News from Seeed Studio","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.seeedstudio.com\/blog\/2020\/06\/04\/nvidia-jetson-nano-and-jetson-xavier-nx-comparison-specifications-benchmarking-container-demos-and-custom-model-inference\/","og_locale":"en_US","og_type":"article","og_title":"NVIDIA Jetson Nano and Jetson Xavier NX Comparison: Specifications, Benchmarking, Container Demos, and Custom Model Inference - Latest News from Seeed Studio","og_description":"New release for NVIDIA Jetson Nano and Jetson Xavier NX reComputer Jetson series are compact","og_url":"https:\/\/www.seeedstudio.com\/blog\/2020\/06\/04\/nvidia-jetson-nano-and-jetson-xavier-nx-comparison-specifications-benchmarking-container-demos-and-custom-model-inference\/","og_site_name":"Latest News from Seeed Studio","article_published_time":"2020-06-04T14:47:41+00:00","article_modified_time":"2022-06-02T03:57:34+00:00","og_image":[{"width":1280,"height":720,"url":"https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2020\/06\/7421591260005_.pic_hd.jpg","type":"image\/jpeg"}],"author":"Elaine Wu","twitter_card":"summary_large_image","twitter_misc":{"Written by":"Elaine Wu","Est. reading time":"13 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/www.seeedstudio.com\/blog\/2020\/06\/04\/nvidia-jetson-nano-and-jetson-xavier-nx-comparison-specifications-benchmarking-container-demos-and-custom-model-inference\/","url":"https:\/\/www.seeedstudio.com\/blog\/2020\/06\/04\/nvidia-jetson-nano-and-jetson-xavier-nx-comparison-specifications-benchmarking-container-demos-and-custom-model-inference\/","name":"NVIDIA Jetson Nano and Jetson Xavier NX Comparison: Specifications, Benchmarking, Container Demos, and Custom Model Inference - Latest News from Seeed Studio","isPartOf":{"@id":"https:\/\/www.seeedstudio.com\/blog\/#website"},"primaryImageOfPage":{"@id":"https:\/\/www.seeedstudio.com\/blog\/2020\/06\/04\/nvidia-jetson-nano-and-jetson-xavier-nx-comparison-specifications-benchmarking-container-demos-and-custom-model-inference\/#primaryimage"},"image":{"@id":"https:\/\/www.seeedstudio.com\/blog\/2020\/06\/04\/nvidia-jetson-nano-and-jetson-xavier-nx-comparison-specifications-benchmarking-container-demos-and-custom-model-inference\/#primaryimage"},"thumbnailUrl":"https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2020\/06\/7421591260005_.pic_hd.jpg","datePublished":"2020-06-04T14:47:41+00:00","dateModified":"2022-06-02T03:57:34+00:00","author":{"@id":"https:\/\/www.seeedstudio.com\/blog\/#\/schema\/person\/61c04bed5bbe2d098f04195c6e48fb11"},"breadcrumb":{"@id":"https:\/\/www.seeedstudio.com\/blog\/2020\/06\/04\/nvidia-jetson-nano-and-jetson-xavier-nx-comparison-specifications-benchmarking-container-demos-and-custom-model-inference\/#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.seeedstudio.com\/blog\/2020\/06\/04\/nvidia-jetson-nano-and-jetson-xavier-nx-comparison-specifications-benchmarking-container-demos-and-custom-model-inference\/"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/www.seeedstudio.com\/blog\/2020\/06\/04\/nvidia-jetson-nano-and-jetson-xavier-nx-comparison-specifications-benchmarking-container-demos-and-custom-model-inference\/#primaryimage","url":"https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2020\/06\/7421591260005_.pic_hd.jpg","contentUrl":"https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2020\/06\/7421591260005_.pic_hd.jpg","width":1280,"height":720,"caption":"NVIDIA Jetson Nano and Jetson Xavier NX Comparison: Specifications, Benchmarking, Container Demos, and Custom Model Inference"},{"@type":"BreadcrumbList","@id":"https:\/\/www.seeedstudio.com\/blog\/2020\/06\/04\/nvidia-jetson-nano-and-jetson-xavier-nx-comparison-specifications-benchmarking-container-demos-and-custom-model-inference\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/www.seeedstudio.com\/blog\/"},{"@type":"ListItem","position":2,"name":"NVIDIA Jetson Nano and Jetson Xavier NX Comparison: Specifications, Benchmarking, Container Demos, and Custom Model Inference"}]},{"@type":"WebSite","@id":"https:\/\/www.seeedstudio.com\/blog\/#website","url":"https:\/\/www.seeedstudio.com\/blog\/","name":"Latest News from Seeed Studio","description":"Emerging IoT, AI and Autonomous Applications on the Edge","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/www.seeedstudio.com\/blog\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"},{"@type":"Person","@id":"https:\/\/www.seeedstudio.com\/blog\/#\/schema\/person\/61c04bed5bbe2d098f04195c6e48fb11","name":"Elaine Wu","image":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/www.seeedstudio.com\/blog\/#\/schema\/person\/image\/","url":"https:\/\/secure.gravatar.com\/avatar\/184af8ef71f0d6b64c276f9bb38b992e?s=96&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/184af8ef71f0d6b64c276f9bb38b992e?s=96&r=g","caption":"Elaine Wu"},"description":"Head of AI Robotics @seeed Every day holds new magic \u2728 on ne sait jamais\u2601\ufe0f","sameAs":["https:\/\/www.linkedin.com\/in\/elaine1994\/"],"url":"https:\/\/www.seeedstudio.com\/blog\/author\/elaine\/"}]}},"modified_by":"Elaine Wu","views":86055,"featured_image_urls":{"full":["https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2020\/06\/7421591260005_.pic_hd.jpg",1280,720,false],"thumbnail":["https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2020\/06\/7421591260005_.pic_hd-80x80.jpg",80,80,true],"medium":["https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2020\/06\/7421591260005_.pic_hd-300x169.jpg",300,169,true],"medium_large":["https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2020\/06\/7421591260005_.pic_hd-768x432.jpg",640,360,true],"large":["https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2020\/06\/7421591260005_.pic_hd-1030x579.jpg",640,360,true],"1536x1536":["https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2020\/06\/7421591260005_.pic_hd.jpg",1280,720,false],"2048x2048":["https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2020\/06\/7421591260005_.pic_hd.jpg",1280,720,false],"visody_icon":["https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2020\/06\/7421591260005_.pic_hd.jpg",32,18,false],"magazine-7-slider-full":["https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2020\/06\/7421591260005_.pic_hd.jpg",1280,720,false],"magazine-7-slider-center":["https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2020\/06\/7421591260005_.pic_hd-936x720.jpg",936,720,true],"magazine-7-featured":["https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2020\/06\/7421591260005_.pic_hd-1024x576.jpg",1024,576,true],"magazine-7-medium":["https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2020\/06\/7421591260005_.pic_hd-720x380.jpg",720,380,true],"magazine-7-medium-square":["https:\/\/www.seeedstudio.com\/blog\/wp-content\/uploads\/2020\/06\/7421591260005_.pic_hd-675x450.jpg",675,450,true]},"author_info":{"display_name":"Elaine Wu","author_link":"https:\/\/www.seeedstudio.com\/blog\/author\/elaine\/"},"category_info":"<a href=\"https:\/\/www.seeedstudio.com\/blog\/category\/news\/\" rel=\"category tag\">News<\/a>","tag_info":"News","comment_count":"0","_links":{"self":[{"href":"https:\/\/www.seeedstudio.com\/blog\/wp-json\/wp\/v2\/posts\/33374","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.seeedstudio.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.seeedstudio.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.seeedstudio.com\/blog\/wp-json\/wp\/v2\/users\/200"}],"replies":[{"embeddable":true,"href":"https:\/\/www.seeedstudio.com\/blog\/wp-json\/wp\/v2\/comments?post=33374"}],"version-history":[{"count":4,"href":"https:\/\/www.seeedstudio.com\/blog\/wp-json\/wp\/v2\/posts\/33374\/revisions"}],"predecessor-version":[{"id":64703,"href":"https:\/\/www.seeedstudio.com\/blog\/wp-json\/wp\/v2\/posts\/33374\/revisions\/64703"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.seeedstudio.com\/blog\/wp-json\/wp\/v2\/media\/33375"}],"wp:attachment":[{"href":"https:\/\/www.seeedstudio.com\/blog\/wp-json\/wp\/v2\/media?parent=33374"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.seeedstudio.com\/blog\/wp-json\/wp\/v2\/categories?post=33374"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.seeedstudio.com\/blog\/wp-json\/wp\/v2\/tags?post=33374"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}