C++ Port of Darknet (of YOLO fame)

OpenCV3 failures when working with C based DL frameworks, like DeepNet (Made famous by YOLO - http://pjreddie.com/darknet/yolo/) is a common issue.

Here is the latest version of Darknet, ported to C++, fixing many coding bugs along the way. Work involved primarily encapsulation of APIs with C linkages, including undefined headers, bug fixes, and typecasting various allocations to actual types, and using correct Error detection types for CUBLAS.

For training with own dataset, and detection, refer to the updated README at,

machine learningyolocaffedarknetc++

Introducing Euclid, a labeller for image-datasets

Introduction:
Euclid is a tool for manual labelling of data - sets, such as those found in Deep learning systems that employ Caffe. This includes systems like DIGITS, and YOLO. It is an object / class labelling tool for machine learning frameworks.

https://github.com/prabindh/euclid/

This tool runs on Linux and Windows, and is based on Python 2.7/Tkinter. Though the UI is a bit clunky it is functional at this point.

Label format support:

digitsdetectnetcaffelabellingeuclid

Impact of Qualcomm-NXP-Freescale on the GPU Ecosystem

Its been a busy week in the markets, with the announcement of the Qualcomm-NXP deal. As is common in such big ticket announcements, a lot of detail gets shoved behind for later discussion.

One of these "little" things is the number of GPU architectures that this combined entity will support at start, and where it will end up later. The same applies to the CPU architectures, but given that it is still coming from an ARM based family it might be less of an issue. On the DSP side, Freescale and Qualcomm also have their own separate DSP engines that need to have their own roadmaps.

gpuadrenovivante

From Zero to Lasagne on Windows 10 (64-bit), on Dell Precision with Quadro GPU, with CUDNN support

From Zero to Lasagne on Windows 10 (64-bit), on Dell Precision with Quadro GPU, with CUDNN support (Replicated from https://github.com/prabindh/Lasagne/wiki/From-Zero-to-Lasagne-on-Windows...(64-bit),-on-Dell-Precision-with-Quadro-GPU,-with-CUDNN-support till the Wiki gets mainlined)

cudacudnnlasagnewindows10theanowinpython

Caffe on Ubuntu 16.04 with VirtualBox + Windows 10 x64 Dell Precision + Quadro 1000

Note: Detailed installation steps of Nvidia DIGITS on Ubuntu 16.04 (native, not VMWare on Windows10), is at https://github.com/prabindh/deepnotes/tree/master/ubuntu16.04.1

Steps

1. Install the AMD64 ISO on the VM using standard procedure

http://releases.ubuntu.com/16.04/ubuntu-16.04.1-desktop-amd64.iso

VMWare version - 5.1.2 r108956

2. sudo apt-get install build-essential

caffemlpythonubuntu 16.04vmwarewindows10quadrocuda

Virtual Reality - 9 issues

Introduction

As VR (Virtual Reality) begins its second (or third) cycle of push into the realms of technology, it is worthwhile to check what problems remain to be solved. This article is a review of past research on fundamental issues that have not yet been solved.

Note: In this article, VR will primarily refer to immersiveness in the Visual domain, and specifically with HMDs (Head Mounted Displays).

Pages