[2017-07-24]increment resources

[ysh329]

madeye/yolo-android: Quantized Tiny Yolo Demo on Android https://github.com/madeye/yolo-android

Analytical Guarantees on Numerical Precision of Deep Neural Networks http://proceedings.mlr.press/v70/sakr17a.html

基于mtcnn和facenet的实时人脸检测与识别系统开发 - 知乎专栏 https://zhuanlan.zhihu.com/p/25025596?refer=shanren7

OAID/mtcnn: C++ project to implement MTCNN, a perfect face detect algorithm, on different DL frameworks. The most popular frameworks: caffe/mxnet/tensorflow, are all suppported now https://github.com/oaid/mtcnn

Seanlinx/mtcnn https://github.com/Seanlinx/mtcnn

foreverYoungGitHub/MTCNN: Repository for "Joint Face Detection and Alignment using Multi-task Cascaded Convolutional Neural Networks", implemented with Caffe, C++ interface. https://github.com/foreverYoungGitHub/MTCNN

CongWeilin/mtcnn-caffe: Joint Face Detection and Alignment using Multi-task Cascaded Convolutional Neural Networks https://github.com/CongWeilin/mtcnn-caffe

[ysh329]

[1703.09746v3] Coordinating Filters for Faster Deep Neural Networks https://arxiv.org/abs/1703.09746v3

[1707.08005v1] Towards Evolutional Compression https://arxiv.org/abs/1707.08005v1

[ysh329]

chuanqi305/MobileNet-SSD: Caffe implementation of Google MobileNet SSD detection network, with pretrained weights on VOC0712 and mAP=0.727. https://github.com/chuanqi305/MobileNet-SSD

christopher5106/FastAnnotationTool: A tool using OpenCV to annotate images for image classification, optical character reading, ... https://github.com/christopher5106/FastAnnotationTool

neubig/lxmls-2017: Slides/code for the Lisbon machine learning school 2017 https://github.com/neubig/lxmls-2017

Squeeze-and-Excitation networks (ILSVRC 2017 winner) at CVPR2017 - Google 照片 https://photos.google.com/share/AF1QipNRXiNDP9tw-B_kyKk4hnXL_N283IaWNxSYH7jtAN1N0m62Uydh3MnpWFPh2GQYUw?key=STNBSU5XRkpKLXBSbmE2Um9GbGRUSm9aME1naFF3

aquaviter/iot-demo-mxnet-greengrass https://github.com/aquaviter/iot-demo-mxnet-greengrass

LaoDar/cnn_head_pose_estimator: a simple and fast mxnet version CNN based head pose estimator https://github.com/LaoDar/cnn_head_pose_estimator

msracver/FCIS: Fully Convolutional Instance-aware Semantic Segmentation https://github.com/msracver/FCIS

naibaf7/libdnn: Greentea LibDNN - a universal convolution implementation supporting CUDA and OpenCL https://github.com/naibaf7/libdnn

blei-lab/edward: A library for probabilistic modeling, inference, and criticism. Deep generative models, variational inference. Runs on TensorFlow. https://github.com/blei-lab/edward

dmlc/nnvm-fusion: Kernel Fusion and Runtime Compilation Based on NNVM https://github.com/dmlc/nnvm-fusion

facebook/fb-caffe-exts: Some handy utility libraries and tools for the Caffe deep learning framework. https://github.com/facebook/fb-caffe-exts

DeepMark/deepmark: THE Deep Learning Benchmarks https://github.com/DeepMark/deepmark

[ysh329]

Lift: Home http://www.lift-project.org/index.html#code

[ysh329]

yonghenglh6/DepthwiseConvolution: A personal mobile convolution implementation on caffe by liuhao.(only GPU) https://github.com/yonghenglh6/DepthwiseConvolution

[ysh329]

ProjectDent/ARKit-CoreLocation: Combines the high accuracy of AR with the scale of GPS data. https://github.com/ProjectDent/ARKit-CoreLocation

[ysh329]

How to squeeze the most from your training data http://www.kdnuggets.com/2017/07/squeeze-most-from-training-data.html

[ysh329]

Pili-完美直播体验（Pili Streaming Cloud） https://github.com/pili-engineering

pili-engineering/PLDroidMediaStreaming: PLDroidMediaStreaming 是 Pili 直播 SDK 的 Android 推流端，支持 RTMP 推流，h.264 和 AAC 编码，硬编、软编支持。具有丰富的数据和状态回调，方便用户根据自己的业务定制化开发。具有直播场景下的重要功能，如：美颜、背景音乐、水印等功能。PLDroidMediaStreaming 是现在目前重点维护的版本，自带采集模块也支持用户自己做采集端。 https://github.com/pili-engineering/PLDroidMediaStreaming

pili-engineering/PLDroidShortVideo: PLDroidShortVideo 是七牛推出的一款适用于 Android 平台的短视频 SDK，提供了包括美颜、滤镜、水印、断点录制、分段回删、视频编辑、混音特效、本地/云端存储在内的多种功能，支持高度定制以及二次开发。 https://github.com/pili-engineering/PLDroidShortVideo

pili-engineering/PLDroidPlayer: PLDroidPlayer 是 Pili 直播 SDK 的安卓播放器。支持所有直播常用的格式，如：RTMP、HLS、FLV。拥有优秀的功能和特性，如：首屏秒开、追帧优化、丰富的数据和状态回调、硬解软解支持。而且可以根据自己的业务进行高度定制化开发。 https://github.com/pili-engineering/PLDroidPlayer

pili-engineering/PLMediaStreamingKit: PLMediaStreamingKit 是 Pili 直播 SDK 的 iOS 推流端，支持 RTMP 推流，h.264 和 AAC 编码，硬编、软编支持。具有丰富的数据和状态回调，方便用户根据自己的业务定制化开发。具有直播场景下的重要功能，如：美颜、背景音乐、水印等功能。 https://github.com/pili-engineering/PLMediaStreamingKit

pili-engineering/PLShortVideoKit: PLShortVideoKit 是七牛推出的一款适用于 iOS 平台的短视频 SDK，提供了包括美颜、滤镜、水印、断点录制、分段回删、视频编辑、混音特效、本地/云端存储在内的多种功能，支持高度定制以及二次开发。 https://github.com/pili-engineering/PLShortVideoKit

pili-engineering/PLPlayerKit: PLPlayerKit 是 Pili 直播 SDK 的 iOS 播放器。支持所有直播常用的格式，如：RTMP、HLS、FLV。拥有优秀的功能和特性，如：首屏秒开、追帧优化、丰富的数据和状态回调、硬解软解支持。而且可以根据自己的业务进行高度定制化开发。 https://github.com/pili-engineering/PLPlayerKit

pili-engineering/PLPlayerKit: PLPlayerKit 是 Pili 直播 SDK 的 iOS 播放器。支持所有直播常用的格式，如：RTMP、HLS、FLV。拥有优秀的功能和特性，如：首屏秒开、追帧优化、丰富的数据和状态回调、硬解软解支持。而且可以根据自己的业务进行高度定制化开发。 https://github.com/pili-engineering/PLPlayerKit

[ysh329]

Creating insanely fast image classifiers with MobileNet in TensorFlow https://hackernoon.com/creating-insanely-fast-image-classifiers-with-mobilenet-in-tensorflow-f030ce0a2991

OpenMined/Syft: Homomorphically Encrypted Deep Learning Library https://github.com/OpenMined/syft

AngusG/tensorflow-xnor-bnn: BinaryNets in TensorFlow with XNOR GEMM op https://github.com/AngusG/tensorflow-xnor-bnn

[ysh329]

ZhishengWang/Embedded-Neural-Network: collection of works aiming at reducing model sizes or the ASIC/FPGA accelerator for machine learning https://github.com/ZhishengWang/Embedded-Neural-Network

[ysh329]

Samsung/iotjs: Platform for Internet of Things with JavaScript http://www.iotjs.net https://github.com/Samsung/iotjs

juefeix/lbcnn.torch: Torch implementation of CVPR'17 - Local Binary Convolutional Neural Networks http://xujuefei.com/lbcnn.html https://github.com/juefeix/lbcnn.torch

• [1604.08772] Towards Conceptual Compression https://arxiv.org/abs/1604.08772

• [1608.02893] Syntactically Informed Text Compression with Recurrent Neural Networks https://arxiv.org/abs/1608.02893

• [1608.05148v2] Full Resolution Image Compression with Recurrent Neural Networks https://arxiv.org/abs/1608.05148v2

• kedartatwawadi/NN_compression https://github.com/kedartatwawadi/NN_compression

[ysh329]

MTG/essentia: C++ library for audio and music analysis, description and synthesis, including Python bindings https://github.com/MTG/essentia

[ysh329]

"TensorFire - runs neural networks in the browser using WebGL" https://tenso.rs/ style-transfer Demo: https://tenso.rs/demos/fast-neural-style/

​​​《Fine-Pruning: Joint Fine-Tuning and Compression of a Convolutional Network with Bayesian Optimization》F Tung, S Muralidharan, G Mori [Simon Fraser University] (2017) ​​​ https://arxiv.org/abs/1707.09102

kingreza/SeeFood: Inspired by HBO's Silicon Valley: SeeFood is an iOS app that uses CoreML to detect various dishes https://github.com/kingreza/SeeFood

farmingyard/caffe-mobilenet: A caffe implementation of mobilenet's depthwise convolution layer. https://github.com/farmingyard/caffe-mobilenet

[ysh329]

【AI让芯片业洗牌】苹果、微软和谷歌挤入赛道，英特尔、英伟达、高通、AMD几家欢乐几家愁 http://mp.weixin.qq.com/s/WlZTXCRy0xGeuJLQMxZGeQ

人工智能和云计算让芯片业洗牌，英特尔成了最大输家 2017-07-31 专注报道AI 量子位 http://mp.weixin.qq.com/s/G_OEZJ0a62TZuMRq5jpXmA

The Rise of AI Is Forcing Google and Microsoft to Become Chipmakers | WIRED https://www.wired.com/story/the-rise-of-ai-is-forcing-google-and-microsoft-to-become-chipmakers/

OpenMined/Syft: Homomorphically Encrypted Deep Learning Library https://github.com/OpenMined/syft

AngusG/tensorflow-xnor-bnn: BinaryNets in TensorFlow with XNOR GEMM op https://github.com/AngusG/tensorflow-xnor-bnn

Qualcomm Snapdragon Neural Processing Engine (NPE) - Qualcomm Developer Network https://developer.qualcomm.com/software/snapdragon-neural-processing-engine

mlmodelzoo.com – deep learning models on mobile http://mlmodelzoo.com/

www.eecs.harvard.edu/~htk/publication/2016-icpr-teerapittayanon-mcdanel-kung.pdf http://www.eecs.harvard.edu/~htk/publication/2016-icpr-teerapittayanon-mcdanel-kung.pdf

[1707.06990] Memory-Efficient Implementation of DenseNets https://arxiv.org/abs/1707.06990

liuzhuang13/DenseNet: Densely Connected Convolutional Networks, In CVPR 2017 (Best Paper Award). https://github.com/liuzhuang13/DenseNet

Caffe2 Bay Area Meetup (5/31/2017) - YouTube https://www.youtube.com/playlist?list=PLD5D5H5YL9SIjxj3IC019AprtgJAjIU3q

如何评价腾讯刚出的ncnn库？ - 知乎 https://www.zhihu.com/question/62871439

[ysh329]

【了不起的芯片】IEEE 盘点27款震撼世界的芯片，你认识哪些？ https://mp.weixin.qq.com/s?timestamp=1501553824&src=3&ver=1&signature=RHizF2rrsg87PJvWK2ae08p9nWR6Syn*wKesA5uhhsbb*l6OStUtpt7QXuYo4IBLM-YZCu*pZnWDq6*R9fMYb37DtJvoQX*MZdY5oVsjtRHQjby5Ktr1wDBunZbGPIl0pNeguh5RMqElNBmBcvTzFaUmm0wSb8xbd316aFtJO-4=

沈向洋宣布微软开发 AI 芯片HPU，剑指英伟达等芯片巨头软肋 https://mp.weixin.qq.com/s?timestamp=1501553824&src=3&ver=1&signature=RHizF2rrsg87PJvWK2ae08p9nWR6Syn*wKesA5uhhsbb*l6OStUtpt7QXuYo4IBLM-YZCu*pZnWDq6*R9fMYb63vSvVPPV-ce8czOmCPE*hytlYj4qpoNFC0Rshq0-tTOduT6boQFGE847hWGQ*n0uA60Y7hEzmS4We6mf1VCkc=

【AI硅脑】超越GPU，FPGA、ASIC和更智能的手机 https://mp.weixin.qq.com/s?timestamp=1501553824&src=3&ver=1&signature=RHizF2rrsg87PJvWK2ae08p9nWR6Syn*wKesA5uhhsbb*l6OStUtpt7QXuYo4IBLM-YZCu*pZnWDq6*R9fMYb0Zgu93l8ms*zs3D7MR8G65EhryJghp*KlNT6ag2E4M-Pec-ZIGFSTABZb0UPRohOtuGigGc1mJP87IO37qBBhA=

[ysh329]

[ASPLOS’17] Neurosurgeon:collaborative intelligence between the cloud and mobile edge Kang et al. http://web.eecs.umich.edu/~jahausw/publications/kang2017neurosurgeon.pdf

如何让深度学习在手机应用上也能加速跑？看完这篇文章你就知道了 https://mp.weixin.qq.com/s?__biz=MzA4NjA4MTkzMw==&mid=2651989741&idx=1&sn=e1fdbf9d23d05089218c1e4d1be8937b&chksm=8428ae0fb35f2719115120bb11e5a307ef42eb4f2677e762b14602c9307dac35e22abbe3e7e9&scene=0#rd

[ysh329]

HandNet - A dataset of depth images of hands http://www.cs.technion.ac.il/~twerd/HandNet/

Darknet with NNPACK NNPACK was used to optimize Darknet without using a GPU. It is useful for embedded devices using ARM CPUs. https://github.com/thomaspark-pkj/darknet-nnpack

[ysh329]

hollance/BNNS-vs-MPSCNN: Compares the speed of Apple's two deep learning frameworks: BNNS and Metal Performance Shaders https://github.com/hollance/BNNS-vs-MPSCNN

hollance/MobileNet-CoreML: The MobileNet neural network using Apple's new CoreML framework https://github.com/hollance/MobileNet-CoreML

hollance/TensorFlow-iOS-Example: Source code for my blog post "Getting started with TensorFlow on iOS" https://github.com/hollance/TensorFlow-iOS-Example

hollance/Forge: A neural network toolkit for Metal https://github.com/hollance/Forge

[ysh329]

Building Cross-Platform CUDA Applications with CMake | Parallel Forall https://devblogs.nvidia.com/parallelforall/building-cuda-applications-cmake/?_lrsc=dca4b9d4-7747-48e0-b9a0-961aba39a657&ncid=so-twi-lt-799

[ysh329]

Ubuntu16.04---腾讯NCNN框架入门到应用 - ruyiwei.cas@gmail.com - CSDN博客 http://blog.csdn.net/Best_Coder/article/details/76201275

[ysh329]

[1608.01409v5] Faster CNNs with Direct Sparse Convolutions and Guided Pruning https://arxiv.org/abs/1608.01409v5

[1606.05316v2] Learning Infinite-Layer Networks: Without the Kernel Trick https://arxiv.org/abs/1606.05316v2

[1707.09422v1] Hyperprofile-based Computation Offloading for Mobile Edge Networks https://arxiv.org/abs/1707.09422v1

[1705.04630v2] Forecasting using incomplete models https://arxiv.org/abs/1705.04630v2

[1707.09068v1] Tartan: Accelerating Fully-Connected and Convolutional Layers in Deep Learning Networks by Exploiting Numerical Precision Variability https://arxiv.org/abs/1707.09068v1

[1707.09926v1] A Framework for Super-Resolution of Scalable Video via Sparse Reconstruction of Residual Frames https://arxiv.org/abs/1707.09926v1

[1707.09855v1] Convolution with Logarithmic Filter Groups for Efficient Shallow CNN https://arxiv.org/abs/1707.09855v1

[1707.09597v1] ScanNet: A Fast and Dense Scanning Framework for Metastatic Breast Cancer Detection from Whole-Slide Images https://arxiv.org/abs/1707.09597v1

[ysh329]

kevinzakka/DenseNet: PyTorch Implementation of "Densely Connected Convolutional Networks" https://github.com/kevinzakka/DenseNet

[1708.00999] Extreme Low Resolution Activity Recognition with Multi-Siamese Embedding Learning https://arxiv.org/abs/1708.00999

[1708.00999] Extreme Low Resolution Activity Recognition with Multi-Siamese Embedding Learning https://arxiv.org/abs/1708.00999

buptwangfei/residual-attention-network: Residual Attention Network for Image Classification https://github.com/buptwangfei/residual-attention-network

Qualcomm Snapdragon Neural Processing Engine (NPE) - Qualcomm Developer Network https://developer.qualcomm.com/software/snapdragon-neural-processing-engine

timvieira/learning-to-prune: Learning to Prune: Exploring the Frontier of Fast and Accurate Parsing code paper

OpenCV 3.3 - OpenCV library http://opencv.org/opencv-3-3.html

[ysh329]

Merged bi-weekly-report for 2017-08-07