ZJU_CVs Blog

CV Learning and Paper Reading.

PAY LESS ATTENTION WITH LIGHTWEIGHT AND DYNAMIC CONVOLUTIONS

1. Introduction lightweight convolutons 轻量级卷积利用深度可分离卷积为原型，通过在通道维度上的共享参数大大减少参数量，降低算法的复杂度（如下图，以$3\times 3$卷积核为例），其思想在于将通道和区域分开考虑 传统卷积参数：$k^2 \times d_{in} ...

Posted by JY on June 18, 2020

CVPR 2020

论文合集

关于往年CV顶会论文（如CVPR 2019、ICCV 2019、ECCV 2018）以及其他优质CV论文和大盘点，详见： https://github.com/amusi/daily-paper-computer-vision CNN 图像分类 目标检测 3D目标检测 视频目标检测 目标跟踪 语义分割 实例分割 全景分割 视频目标分割 超像素分...

Posted by JY on June 18, 2020

Set Transformer

A Framework for Attention-based Permutation-Invariant Neural Networks

1. Introduction NLP的输入存在有序性，但是还有很多机器学习任务的输入与顺序无关（set-strcture data），如： multiple instance learning 3D shape recognition sequence ordering meta-learning ...

Posted by JY on June 14, 2020

SMASH

one-shot model architecture search through hypernetworks

1. Introduction 为深度神经网络设计架构需要专业知识和大量计算时间（expert knowledge and substantial computation time） 本文通过训练一个辅助模型（HyperNet），去训练搜索过程中的候选模型，这个超网络动态地生成具有可变结构的主模型的权值（generate the weights o f a main mod...

Posted by JY on June 11, 2020

SOLO Segmenting Objects by Locations

SOLO分割模型

1. Introduction 实例分割主要有两大类别： 自上而下：先检测后分割（依赖目标检测的准确率） 自下而上：为每个像素学习一个嵌入向量，然后通过聚类、度量学习等方法区分不同的实例（依赖每个像素的嵌入学习和分组后处理） 创新点：利用中心点和大小两个信息区别实例 2. SOLOv1 SOLO: Segmenting Objects b...

Posted by JY on June 9, 2020

Awesome Self-Supervised Learning

Why Self-Supervised? Self-Supervised Learning has become an exciting direction in AI community. Jitendra Malik: “Supervision is the opium of the AI researcher” Alyosha Efros: “The AI revolutio...

Posted by JY on June 4, 2020

Awesome Image Classification

Background I believe image classification is a great start point before diving into other computer vision fields, espacially for begginers who know nothing about deep learning. When I started to l...

Posted by JY on June 4, 2020

Awesome Few-shot Learning

Meta-learning Model optimzation Unsupervised Meta-Learning for Few-Shot Image and Video Classification [Khodadadeh et al. 2018] A Simple Neural Attentive Meta-Learner [Mishra et al. 2018] Ne...

Posted by JY on June 4, 2020

PFLD A Practical Facial Landmark Detector

人脸对齐

1. Introduction 人脸关键点检测器实际应用所需特征: 准确性好，高效，模型轻量级； 本文提出具有高精度的轻量级landmark检测模型，在移动设备上能达到实时的性能 2. Method 人脸关键点检测难点 局部变化: 现实场景中人脸的表情、光照、以及遮挡情况变化都很大 全局变化: 姿态和成像质量影响图像中人脸的表征，人脸全局结构的错误估计直接导致定...

Posted by JY on June 2, 2020

End-to-End Object Detection with Transformers

目标检测算法

1. Introduction 近年来，Transformer 成为了深度学习领域非常受欢迎的一种架构，它依赖注意力机制，使得 AI 模型有选择地聚焦于输入的某些部分，因此推理更加高效。Transformer 已经广泛应用于序列数据的处理，尤其是在语言建模、机器翻译等自然语言处理领域。此外，它在语音识别、符号数学、强化学习等多个领域也有应用。（Transformer的介绍详...

Posted by JY on May 31, 2020

FEATURED TAGS
Introduction FSL Transfer Learning paper GAN Application Datasets Classification Attention Mechanism Learning Paradigm Anomaly Detection Backbone Cluster Overview Segmentation Model Compression NAS Object Detection Record Facial Tools Transformer SSL ZSL GCN Fine-grained Meta-Learning Data Imputation 开发 架构 Video Analysis Object Tracking interview
ABOUT ME

✉️ 1067087283@qq.com

FRIENDS