- Machine Learning expertise: Google is a dominant force in machine learning. Its prominence in search owes a lot to the strides it achieved in machine learning.
- Scalability: the announcement noted that TensorFlow was initially designed for internal use and that it’s already in production for some live product features.
- Ability to run on Mobile.
This last reason is the operating reason for this post since we’ll be focusing on Android. If you examine the tensorflow repo on GitHub, you’ll find a little tensorflow/examples/android directory. I’ll try to shed some light on the Android TensorFlow example and some of the things going on under the hood.
The nightly builds for Android x86 (Marshmallow) are back for a while now and it seems they’re working as expected.
The bootable ISOs can be downloaded from this link for both 32bit and 64bit architectures. The build is made each night, freshly from the repository and unless the build fails, the ISO-s are updated.
It should boot either from USB or any other media. The page contains information about how to test it.
In some previous topics (here and here) I wrote about some cheap development boards which can be acquired from EBay or Aliexpress. Since System Workbench for STM32 is freely available for a while now, let’s see how can we use it to generate a project, compile it, upload it to a board and debugging it step by step. We’ll use for this the board I got from EBay, but it works the same with the any STM32 other board I have and also with some self-made ones.
For being able to install firmware on the board and debug it, first we need to have a hardware part which will sit between the computer and the board. There are various models and versions of these jtag debugers and they can be ordered online or found pretty cheap on ebay (clones). Another way to get hold of one of these is to have a development board which comes equiped with JTAG adapters, like the STM32 discovery series of boards. Some of these JTAG debuggers allow even breaking apart the JTAG debugger from the development board itself (LPCXpresso series, the nucleo boards).
Regardless of which JTAG interface is used, it should be one which is known to work with OpenOCD, as we’ll use OpenOCD for debugging. In our case we’ll use the stm32f4 discovery board’s stlink2 side. However, Before using it as a JTAG debugger, we need to disconnect the STLink part from the discovery board, by removing two jumpers. Once that is done, the STLink itself won’t be connected to the discovery board and it’s SWD header can be connected to any other board. Read the rest of this entry →
Unreal Engine marketplace started to release code plug-ins. For start, 5 free plug-ins will assist you in gathering analytics on player data, provide support for REST server communications, allow you to load MODO material .xml files for 3D meshes and more.
The Unreal Material Importer is a plug-in for Unreal Engine 4 that you can use to load MODO material .xml files and apply them to 3D meshes of a game level in the Unreal Engine 4 editor.
If you have a MODO scene, you can export 3D meshes in the form of .fbx files, and materials and textures as .xml files.
You can then apply the exported materials to the .fbx 3D meshes in Unreal Engine, using the Unreal Material Importer plug-in.
SkookumScript’s cutting-edge command console turbocharges your workflow (at any stage of development) by enabling you to query and manipulate any UE4 game as it runs on any platform—without disrupting your existing tools, C++ code or BP graphs. So even if you aren’t looking for a scripting solution now, try our console. You’ll love it. We promise.
Then there’s SkookumScript itself—a text-based, compiled language that is made for games. With key game concepts such as concurrency built-in, SkookumScript empowers the entire team—from light coders to C++ veterans—to create sophisticated gameplay with surprisingly few lines of code. Its addictively useful IDE features live code changes with instant turnaround, context-sharing with the UE4 editor, and remote debugging. It painlessly scales with team size and content, and benevolently bridges between C++ and Blueprints—changes in SkookumScript are reflected live in Blueprint graphs and vice versa. Wow!
Lovingly crafted by veteran game developers, battle-tested on the hit AAA titles “Sleeping Dogs” and “Sleeping Dogs: Definitive Edition”, and now in use on several upcoming AAA and indie games, SkookumScript fills your game development experience with cackles of megalomaniacal glee. Better coding through mad science!
Understand your players’ in-game behaviour with the free GameAnalytics Plugin.
GameAnalytics collect player data and provides a powerful set of features that enables you to analyse in-game behaviour.
Start your analysis within 5 minutes, with our range of predefined dashboards (Real-time, Acquisition, Engagement, 1st monetizers, Monetization, Progression, Resources).
Track, visualize and evaluate:
• Player progression – Balance your levels and find out where your players struggle;
• In-game economy – Measure what your players are sinking their gold on and more;
• Custom dimensions – Track any relevant interaction with your game;
• Real money transactions with purchases validation – Analyze your validated revenue on all IAP purchases;
• Funnels – Improve your game, by digging into any sequence of events, by any segmentation;
• Error tracking – Investigate the quality of your game;
• and much more…
GameAnalytics makes it easy to assess your game mechanics, design and economy with the reports and tools it provides.
This runtime plugin links to our .libs which find and load our SDK .dlls shipped with Logitech Gaming Software. LGS and the SDK .dlls are responsible for all the work done to communicate to each of the devices. We designed this loading scheme to allow for older versions of our .lib to have support for newer devices by upgrading LGS. This takes the burden off the game developer by shipping a smaller .lib and also ensures future support.
This plugin enables the control of Logitech Gaming products by porting these SDKs to the UE4 engine:
• Logitech|G ARX Control SDK
• Logitech|G LED Illumination SDK
• Logitech|G G-Key Macro SDK
• Logitech|G LCD Gamepanel SDK
• Logitech|G Steering Wheel SDK
VaRest is the plugin for Unreal Engine 4 that makes REST server communications easier to use.
List of Modules:
• VaRestPlugin (Runtime)
• VaRestEditorPlugin (Editor)
List of Features:
• Flexible Http/Https request management with support of different Verbs and Content Types
• No C++ coding required, everything can be managed via blueprints
• Blueprintable FJsonObject wrapper with almost full support of Json features: different types of values, arrays, binary data content, both ways serializarion to FString, etc.
• Blueprintable FJsonValue wrapper – full Json features made for blueprints!
• Both bindable events and latent functions are provided to control the asynchronous requests
Source: Unreal Engine blog.
Recently I was looking for some cheap boards which would allow me to play with the STM32F103 microcontroller and I’ve found that Aliexpress has couple of versions of them sold quite cheap by various sellers, with a very affordable price tag. There are various incarnations of these boards, and with difference in the price range, so I went with these two in the end:
TCS3771 and alike are a range of I2C RGB sensors allowing one to read not only light intensity but also it’s color. With a bit of care and consideration, the light intensity can be calculated with quite a precision. They provides red, green, blue, and clear (RGBC) light sensing and proximity detection (when coupled with an external IR LED). They detect light intensity under a variety of lighting conditions and through a variety of attenuation materials.
The device contains a 4 × 4 photodiode array, integrating amplifiers, ADCs, accumulators, clocks, buffers, comparators, a state machine, and an I2C interface. The 4 × 4 photodiode array is composed of red-filtered, green-filtered, blue-filtered, and clear photodiodes – four of each type. Four integrating ADCs simultaneously convert the amplified photodiode currents to a digital value providing up to 16 bits of resolution. Upon completion of the conversion cycle, the conversion result is transferred to the data registers. The transfers are double-buffered to ensure that the integrity of the data is maintained. Communication to the device is accomplished through a fast (up to 400kHz), two-wire I2C serial bus for easy connection to a microcontroller or embedded controller.
This article hooks up a TCS3771 to LPC1114 and provides some explanation and code to read the RGB and C values from the device.
Atmel Corporation has announced the world’s lowest power Bluetooth Low Energy (BLE) connectivity modules for small form factor applications. The ultra-low power Atmel SmartConnect XR and ZR modules consume less than 4mA in RX and less than 3mA in TX with 3.6V, and below 1.2µA in sleep mode, increasing battery life by up to 3X longer than other solutions on the market today.
Pushing the limits of space-constrained areas for tomorrow’s Bluetooth applications, the Atmel SmartConnect BTLC1000-XR and SAMB11-XR modules come in an extremely small 4.5×5.5mm LGA package with 40L, while the BTLC1000-ZR and SAMB11-ZR modules come in a 7.5×10.5mm package with 34-pins and a ground paddle. These new modules are ideal for space-constrained Internet of Things (IoT) applications including beacons, sensor tags door locks, wearables and much, much more.
For the entire article, follow this link.
Android Studio 2.0 introduces a faster emulator, new Instant Run, app indexing, and Cloud Test Lab features, and an improved GPU Developer debugger.