晓飞的算法工程笔记

业余时间更新一些算法和工程笔记,更多内容,请关注公众号【晓飞的算法工程笔记】
69 篇文章
13 人订阅

全部文章

VincentLee

DDBNet:Anchor-free新训练方法,边粒度IoU计算以及更准确的正负样本 | ECCV 2020

论文: Dive Deeper Into Box for Object Detection

10130
VincentLee

Dynamic ReLU:微软推出提点神器,可能是最好的ReLU改进 | ECCV 2020

ReLU是深度学习中很重要的里程碑,简单但强大,能够极大地提升神经网络的性能。目前也有很多ReLU的改进版,比如Leaky ReLU和 PReLU,而这些改进版...

6100
VincentLee

APReLU:跨界应用,用于机器故障检测的自适应ReLU | IEEE TIE 2020

论文: Deep Residual Networks with Adaptively Parametric Rectifier Linear Units for...

9300
VincentLee

AABO:自适应Anchor设置优化,性能榨取的最后一步 | ECCV 2020 Spotlight

论文: AABO: Adaptive Anchor Box Optimization for Object Detection via Bayesian Sub...

8600
VincentLee

CSG:清华大学提出通过分化类特定卷积核来训练可解释的卷积网络 | ECCV 2020 Oral

论文: Training Interpretable Convolutional Neural Networks by Differentiating Clas...

10830
VincentLee

PIoU Loss:倾斜目标检测专用损失函数,公开数据集Retail50K | ECCV 2020 Spotlight

论文: PIoU Loss: Towards Accurate Oriented Object Detection in Complex Environment...

11510
VincentLee

简单的特征值梯度剪枝,CPU和ARM上带来4-5倍的训练加速 | ECCV 2020

**论文: Accelerating CNN Training by Pruning

6320
VincentLee

Jigsaw pre-training:摆脱ImageNet,拼图式主干网络预训练方法 | ECCV 2020

论文: Cheaper Pre-training Lunch: An Efficient Paradigm for Object Detection

8110
VincentLee

DGC:真动态分组卷积,可能是解决分组特征阻塞的最好方案 | ECCV 2020 Spotlight

论文: Dynamic Group Convolution for Accelerating Convolutional Neural Networks

9300
VincentLee

S2DNAS:北大提出动态推理网络搜索,加速推理,可转换任意网络 | ECCV 2020 Oral

论文: S2DNAS: Transforming Static CNN Model for Dynamic Inference via Neural Archi...

9720
VincentLee

Gradient Centralization: 一行代码加速训练并提升泛化能力 | ECCV 2020 Oral

论文: Gradient Centralization: A New Optimization Technique for Deep Neural Networ...

11830
VincentLee

DETR:基于Transformer的目标检测新范式,性能媲美Faster RCNN | ECCV 2020 Oral

论文: End-to-End Object Detection with Transformers

14120
VincentLee

BorderDet:通过边界特征大幅提升检测准确率,即插即用且速度不慢 | ECCV 2020 Oral

论文: BorderDet: Border Feature for Dense Object Detection

11130
VincentLee

Gradient Centralization: 一行代码加速训练并提升泛化能力 | ECCV 2020 Oral

论文: Gradient Centralization: A New Optimization Technique for Deep Neural Networ...

10420
VincentLee

CondenseNet:可学习分组卷积,原作对DenseNet的轻量化改造 | CVPR 2018

论文:Neural Architecture Search with Reinforcement Learning

12251
VincentLee

MnasNet:经典轻量级神经网络搜索方法 | CVPR 2019

论文: MnasNet: Platform-Aware Neural Architecture Search for Mobile

12620
VincentLee

MobileNetV1/V2/V3简述 | 轻量级网络

论文: MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applic...

13930
VincentLee

ShuffleNetV1/V2简述 | 轻量级网络

论文: ShuffleNet: An Extremely Efficient Convolutional Neural Network for Mobile D...

15230
VincentLee

SqueezeNet/SqueezeNext简述 | 轻量级网络

论文: SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MB mode...

9330
VincentLee

基于层级表达的高效网络搜索方法 | ICLR 2018

论文: Hierarchical Representations for Efficient Architecture Search

9720

扫码关注云+社区

领取腾讯云代金券