2017年11月_fourye007

11月 10月 09月 08月 07月 06月 05月 04月 03月 02月 01月

原创分类Global-View

Fine-Grained recognitionrequire recognition of highly localized attributes of objects while being invariant to their pose and location in the imagePart-based models construct representations by loca

2017-11-20 09:58:46 495

原创 Highly Efficient Forward and Backward Propagation of Convolutional Neural Networks for Pixelwise Cla

Contributionseliminate all the redundant computation in convolution and pooling on images by introducing novel d-regularly sparse kernels.It generates exactly the same results as those by patch-by-pa

2017-11-06 16:24:24 504

原创 SQUEEZENET

Compelling AdvantagesSmaller CNNs require less communication across servers during distributed training.Smaller CNNs require less bandwidth to export a new model from the cloud to an autonomous car.

2017-11-05 10:32:17 548

原创 DenseNet--Densely Connected Convolutional Networks

Compelling advantagesalleviate the vanishing-gradient problemstrengthen feature propagationencourage feature reusesubstantially reduce the number of parameters.requires fewer parametersno need to

2017-11-03 22:16:33 415

原创 ResNet--Deep Residual Learning for Image Recognition

Key questionVanishing/exploding gradients hamper convergence from the beginning, as the network becomes more deeper.with the network depth increasing, accuracy gets saturated (which might be unsurpri

2017-11-03 20:57:57 326

原创 VGG--VERY DEEP CONVOLUTIONAL NETWORKS FOR LARGE-SCALE IMAGE RECOGNITION

Key Pointswe fix other parameters of the architecture, and steadily increase the depth of the network by adding more convolutional layers, which is feasible due to the use of very small (3 × 3) convol

2017-11-03 19:47:13 271

原创 Maxout Networks

Motivationin multiple dimensions a maxout unit can approximate arbitrary convex functionsContributionsmaxout is cross channel poolingmaxout enhances dropout’s abilities as a model averaging techniq

2017-11-03 19:35:28 419

原创 Network in Network

Key ProblemsCNN implicitly makes the assumption that the latent concepts are linearly separablethe data for the same concept often live on a nonlinear manifold, therefore the representations that cap

2017-11-03 17:17:07 267

原创 AlexNet--ImageNet Classification with Deep Convolutional Neural Networks

LRN(Local Response Normalization)applied this normalization after applying the ReLU nonlinearity in certain layersOverlapping PoolingWe generally observe during training that models with overlapping

2017-11-03 16:48:27 334

opencv缺失文件

使用Cmake生成OpenCV3.2文件时会报缺少部分文件： boostdesc_bgm_bi.i boostdesc_bgm_hd.i boostdesc_bgm.i boostdesc_binboost_256.i boostdesc_binboost_128.i boostdesc_binboost_064.i vgg_generated_120.i vgg_generated_80.i boostdesc_lbgm.i ippicv_linux_20151201.tgz vgg_generated_64.i vgg_generated_48.i protobuf-cpp-3.1.0.tar.gz

2017-07-25