Tag:convolution

  • Hengyuan cloud_ Notes on [image smoothing]

    Time:2022-5-6

    Source of the article | Hengyuan cloud community (a shared computing platform focusing on AI industry:Hengyuan zhixiangyun) Original address|Image smoothing Original author | inster Learning objectives Understand the types of noise in the imageUnderstand the content of average filtering, Gaussian filtering, median filtering, etcBe able to process images using filters 1 image noise Because the […]

  • Lightweight Network Overview – backbone network

    Time:2022-5-5

    The core of lightweight network is the lightweight transformation of the network from both volume and speed on the premise of maintaining the accuracy as much as possible. This paper briefly describes the lightweight network, mainly involving the following networks: Squeezenet series Shufflenet series MnasNet Mobilenet series CondenseNet Espnet series ChannelNets PeleeNet IGC series Fbnet […]

  • Hire MLP of vision MLP: vision MLP via hierarchical arrangement

    Time:2022-5-4

    Hire-MLP: Vision MLP via Hierarchical Rearrangement Original document:https://www.yuque.com/lart/pa… This article is very easy to read. There are no complex words and sentence patterns. It is very smooth from beginning to end. I like this writing style very much. Learn about the article from the summary This paper presents Hire-MLP, a simple yet competitive vision MLP […]

  • TF learning convolutional neural networks convolutional network model

    Time:2022-5-2

    Convolutional neural networks Convolution operation and pooling operationDeep separable convolutionData enhancementTransfer learning General structure: Convolutional neural network (convolution+subsampling)n + fully connected layers,FCm classification Full convolution neural network (convolution+subsampling)n + pixelwise classificationK Object segment Convolution operation:Solve the problem of too many neural network parameters, for example, the picture size is 10001000, the next layer neuron is […]

  • Interpretation of five mainstream deep network models in the industry

    Time:2022-5-1

    Summary:This paper introduces the main methods of model optimization in the industry, and then focuses on model quantification, including the basic principle of quantification, method classification, future development, and interpretation of cutting-edge papers. This article is shared from Huawei cloud community《Summary and application of model quantification》, by Alan_ wen。 preface With the continuous development of […]

  • Raftmlp do MLP based models dream of winning over CV?

    Time:2022-5-1

    RaftMLP: Do MLP-based Models Dream of Winning Over Computer Vision? Original document:https://www.yuque.com/lart/pa… Understanding papers from abstracts For the past ten years, CNN has reigned supreme in the world of computer vision, but recently, Transformer is on the rise. However, the quadratic computational cost of self-attention has become a severe problem of practice. It is pointed […]

  • Cycle MLP a MLP like architecture for dense prediction of visoin MLP

    Time:2022-4-22

    Cycle MLP a MLP like architecture for dense prediction of visoin MLP Original document:https://www.yuque.com/lart/pa… Read the article from the abstract This paper presents a simple MLP-like architecture, CycleMLP, which is a versatile backbone for visual recognition and dense predictions, unlike modern MLP architectures, e.g., MLP-Mixer, ResMLP, and gMLP, whose architectures are correlated to image size […]

  • AlexNet

    Time:2022-4-21

    catalogue Features of alexnet Using relu activation function to accelerate convergence How to understand the nonlinearity of relu function Paper: ImageNet Classification with Deep Convolutional Neural Networks Github:https://github.com/pytorch/vision/blob/master/torchvision/models/alexnet.py AlexNetThe model has the University of Toronto,Geoff HintonLaboratory design, won2012yearImageNet ILSVRCThe champion of the competition, and the error rate is far lower than the second, which makes […]

  • Share real records | use megengine distributed communication operator to realize complex parallel training

    Time:2022-4-21

    In the megengine meetup on March 25th, lecturer Zhou Yizhuang of Kuangshi Research Institute shared “using megengine distributed communication operator to realize complex parallel training”. Live playback link:Using megengine’s distributed communication operator to realize complex parallel training – megengine meetup No.2_ Beeping beeping beeping beeping beeping beeping beeping beeping beeping beeping beeping beeping beeping beeping […]

  • Real time facial expression detection using convolutional neural network

    Time:2022-4-11

    In social interaction, facial expression plays a vital role in nonverbal communication. Psychologist Paul Ekman suggested that people all over the world have seven emotional expressions: happiness, sadness, surprise, fear, anger, disgust and contempt. Building better human-computer interaction, such as detecting human emotions through images, may be a difficult task. Facial expressions are important for […]

  • [let’s fight online with Xiao Mi!] Image classification using mobilenetv2

    Time:2022-4-6

    brief introduction At present, neural network models emerge in endlessly, which is not only very efficient and fast, but also very accurate in the field of image recognition. However, we will never stop on the road of improving the accuracy. Paradoxically, improving the accuracy will also bring consumption and require higher computing resources, which is […]

  • Technology blog – neural network is no longer “rolled” and depends on “Transformers”

    Time:2022-4-4

    author: TA Ying Cheng, Ph.D. candidate of Oxford University and blogger of medium technology. Many articles have been included in the official journal of the platform, towards data sciencePicture source:Unsplash At present, convolutional neural network (CNN) has become the main technical pillar of depth network in computer vision and image related tasks. Compared with the […]