Changes between Version 2 and Version 3 of TPU
- Timestamp:
- 06/18/2020 11:49:02 PM (5 years ago)
Legend:
- Unmodified
- Added
- Removed
- Modified
-
TPU
v2 v3 3 3 The Coral Edge TPU provides a means to perform advanced machine learning tasks in a low power, small form factor package. This hardware is based on an application specific IC (ASIC) tailored for hardware accelerated AI calculations. 4 4 5 This TPU is appropr eate for an application where identifying an object or pattern is required. This could be, but is not limited to:5 This TPU is appropriate for an application where identifying an object or pattern is required. This could be, but is not limited to: 6 6 * Object detection 7 7 * Pose or gesture estimation … … 13 13 * Robots 14 14 * Voice control/Language Processing 15 * Monit ering devices16 17 Nearly all industries can ben ifit from this technology. To name a few more specifically:15 * Monitoring devices 16 17 Nearly all industries can benefit from this technology. To name a few more specifically: 18 18 * Health Care 19 19 * Agrictuture … … 23 23 * Automated Kiosk 24 24 25 For more information about the hardware see these ref rences:25 For more information about the hardware see these references: 26 26 * https://coral.ai 27 27 * https://en.wikipedia.org/wiki/Tensor_processing_unit … … 40 40 == Compiling the kernel 41 41 42 The Gateworks kernel defconfig for Newport does not include support for video devices. For the sake of convenience a pre-built image is avai able for download. If you would like to create a similar image manually:42 The Gateworks kernel defconfig for Newport does not include support for video devices. For the sake of convenience a pre-built image is available for download. If you would like to create a similar image manually: 43 43 44 44 * Acquire the [wiki:/newport/bsp Newport BSP], we will call the directory this repo has been sync'ed to the <BSP> directory. 45 45 * Follow the steps [wiki:/newport/bsp#Modifyingthestand-aloneLinuxKernelieforUbuntu here] to modify the kernel and create and Ubuntu image. 46 46 * In the menuconfig enable the module "USB_VIDEO_CLASS", this will allow you to use a USB webcam with v4l-utils. 47 * Complete the procedure detailed in the afor mentioned section, build your Bionic image and [wiki:/newport/firmware#UpdateFirmwareviaSerialConsoleandEthernetfromBootloader flash it to your SBC]47 * Complete the procedure detailed in the aforementioned section, build your Bionic image and [wiki:/newport/firmware#UpdateFirmwareviaSerialConsoleandEthernetfromBootloader flash it to your SBC] 48 48 49 49 == Building and installing the Gasket and Apex modules … … 52 52 The source code for the modules can be downloaded here as a tar.gz file: 53 53 * https://coral.googlesource.com/linux-imx/+/refs/heads/release-day/drivers/staging/gasket 54 * Extract this tar into a directory, this dir will be refer ed to as <the_module_directory>.54 * Extract this tar into a directory, this dir will be referred to as <the_module_directory>. 55 55 Build the source using the same method as you would for an out of tree module (out of tree = when the module source is not included in the kernel source). 56 56 * cd to the <BSP> directory. … … 75 75 make -C <BSP>/linux M=$PWD 76 76 }}} 77 This procedure will result in two modu eles being created, "apex.ko" and "gasket.ko". Copy the .ko files to your target board "/lib/modules/<kernel_version>/extra/" folder. Using SCP may be the simplest way to go about this.77 This procedure will result in two modules being created, "apex.ko" and "gasket.ko". Copy the .ko files to your target board "/lib/modules/<kernel_version>/extra/" folder. Using SCP may be the simplest way to go about this. 78 78 79 79 With the modules copied to the board execute the following commands: … … 89 89 On reboot verify that "/dev/apex_0" device is present. 90 90 91 == Installing and configur eing Python91 == Installing and configuring Python 92 92 93 93 Python 3.7 is required to run the tenser flow examples. Other versions can be used, though at the time of writing this wiki 3.7 is the best option when using the Bionic Ubuntu for Newport BSP. … … 134 134 }}} 135 135 136 ***Note:*** If you have chosen not to use Python 3.7 you can find the .whl appropr eate for your version here https://www.tensorflow.org/lite/guide/python136 ***Note:*** If you have chosen not to use Python 3.7 you can find the .whl appropriate for your version here https://www.tensorflow.org/lite/guide/python 137 137 138 138 Acquire necessary libraries to build the TFlite runtime … … 149 149 150 150 151 == Download classifcation example, test an infrencingoperation151 == Download Classification example, test an inference operation 152 152 153 153 Create a place for the Coral examples to reside. … … 186 186 git clone https://github.com/google-coral/examples-camera.git --depth 1 187 187 }}} 188 Download models. Models are the information that will be feed to the TPU for it to ref rence when identifying an object.188 Download models. Models are the information that will be feed to the TPU for it to reference when identifying an object. 189 189 {{{#!bash 190 190 cd examples-camera … … 196 196 bash install_requirements.sh 197 197 }}} 198 Adaptations to the existing gstreamer.py will be required for this example to work. A modified script is available to download from [http://dev.gateworks.com/fae/gstreamer.py here].198 Adaptations to the existing gstreamer.py will be required for this example to work. A modified script is available to download from [http://dev.gateworks.com/fae/gstreamer.py here]. 199 199 Edit this file line 231 with the IP address of the desktop workstation you will be streaming to. 200 200 This is what the edited line will look like.