Compact platforms for computer vision and AI computations

-
Currently there are two main brands for compact AI: Intel and NVidia.

Intel Myriad

"Intel's Myriad™ X VPU features a fully tune-able ISP pipeline for the most demanding image and video applications. The Intel® Movidius™ Myriad™ X VPU also features hardware based encode for up to 4K video resolution, meaning the VPU is a single-chip solution for all imaging, computer vision and CNN workloads."

Particular products:

NVidia Jetson

"NVIDIA® Jetson™ systems provide the performance and power efficiency to run autonomous machines software, faster and with less power. Each is a complete System-on-Module (SOM), with CPU, GPU, PMIC, DRAM, and flash storage—saving development time and money."

Products:
  • Jetson AGX Xavier
  • Jetson TX2
  • Jetson Nano




Comments

Post a Comment

Popular posts from this blog

Computing ray origin and direction from Model View Projection matrices for raymarching

-
When performing raymarching (as well a raycasting and raytracing) using fragment shaders in OpenGL, it's required to compute ray's origin and direction. Here we will give solution for finding it from the Model View Projection matrices in OpenGL, implemented in openFrameworks. Using it, you can combine raymarching/raycasting/raytracing with forward OpenGL rendering, so obtain hybrid rendering modes. Also, you can use it for creating VE applications based on raymarching or hybrid rendering. (Actually, we checked what this approach is corrent by creating VR rendering app in HTC Vive). The algorithm can be used on any OpenGL platform with any language supporting matrix inversion computation (in our case it's C++ with GLM library, included in openFrameworks 0.10.1). The approach is based on this hint by  GClements in this discussion on the topic: https://community.khronos.org/t/ray-origin-through-view-and-projection-matrices/7257
Read more

Forward and backward alpha blending for raymarching

-
When implementing raymatching by scanning a ray to accumulating scene's colors, containing semi-transparent colors (for example, scene with clouds), it's required to properly combine colors using alpha blending method. Here we discuss how to do it and obtain corresponding formula. (So, here we propose that objects has "uniform" ambient lighting and some transparency and all we need to accumulate their colors). The forward alpha blending formula, used "everywhere" - in OpenGL, VJ software, photo and video editors, can be formulated in the following way: Formula for "forward" alpha blending ---------------------------------- Let C is a current RGB color, c is a new RGB color and alpha is a new color's opaqueness (alpha is a float value from 0 to 1). Then updating rule for C is the following:     C = (1- alpha )* C + alpha * c    ---------------------------------- So, the more alpha , the more c affec
Read more

Forward, Deferred and Raytracing rendering in openFrameworks and web

-
Image
Currently, there are three basic approaches for rendering computer graphics scenes: forward rendering, deferred rendering, raytracing. 1. Forward rendering It's that OpenGL does, you create vertices, then make faces, and rasterize them, all passing via shaders. The rendering process is straightworward and fast, but rendering shadows, making degree-of-field effect and drawing clouds is hard here. Most of openFrameworks examples use forward rendering. 2. Deferred rendering It's the same as forward rendering, but additionally some extra data is rendered for computing shadows. The lighting pass is performed after the scene geometry is drawn. See detailed explanation here: https://learnopengl.com/Advanced-Lighting/Deferred-Shading Shadows, degree-of-field effect and some kind of reflections are possible with deferred rendering. Unreal Engine 4 and Unity currently (2019) are based on deferred rendering. openFrameworks: https://github.com/perevalovds
Read more