## AlexNet

[【第61篇】AlexNet：CNN開山之作](https://wanghao.blog.csdn.net/article/details/128503264)

## VGGNet

[【第1篇】VGG](https://wanghao.blog.csdn.net/article/details/120094424)

## GooLeNet系列

[【第2篇】GooLeNet](https://wanghao.blog.csdn.net/article/details/120110029)

[【第3篇】Inception V2](https://wanghao.blog.csdn.net/article/details/120148613)

[【第4篇】Inception V3](https://wanghao.blog.csdn.net/article/details/120156107)

[【第62篇】Inception-v4](https://wanghao.blog.csdn.net/article/details/128522765)

## ResNet

[【第5篇】ResNet](https://wanghao.blog.csdn.net/article/details/120178266)

## DenseNet

[【第10篇】DenseNet](https://wanghao.blog.csdn.net/article/details/120347118)

## Swin Transformer

[【第16篇】Swin Transformer](https://wanghao.blog.csdn.net/article/details/120724040)

[【第49篇】Swin Transformer V2：擴(kuò)展容量和分辨率](https://wanghao.blog.csdn.net/article/details/127135297)

## MAE

[【第21篇】MAE（屏蔽自編碼器是可擴(kuò)展的視覺學(xué)習(xí)器）](https://wanghao.blog.csdn.net/article/details/121605608)

## CoAtNet

[【第22篇】CoAtNet：將卷積和注意力結(jié)合到所有數(shù)據(jù)大小上](https://wanghao.blog.csdn.net/article/details/121993729)

## ConvNeXtV1、V2

[【第25篇】力壓Tramsformer，ConvNeXt成了CNN的希望](https://wanghao.blog.csdn.net/article/details/122451111)

[【第64篇】ConvNeXt V2論文翻譯：ConvNeXt V2與MAE激情碰撞](https://wanghao.blog.csdn.net/article/details/128541957?spm=1001.2014.3001.5502)

## MobileNet系列

[【第26篇】MobileNets：用于移動視覺應(yīng)用的高效卷積神經(jīng)網(wǎng)絡(luò)](https://wanghao.blog.csdn.net/article/details/122692846)

[【第27篇】MobileNetV2：倒置殘差和線性瓶頸](https://wanghao.blog.csdn.net/article/details/122729844)

[【第28篇】搜索 MobileNetV3](https://wanghao.blog.csdn.net/article/details/122779006)

## MPViT

[【第29篇】MPViT：用于密集預(yù)測的多路徑視覺轉(zhuǎn)換器](https://wanghao.blog.csdn.net/article/details/122782937)

## VIT

[【第30篇】Vision Transformer](https://wanghao.blog.csdn.net/article/details/123695223)

## SWA

[【第32篇】SWA：平均權(quán)重導(dǎo)致更廣泛的最優(yōu)和更好的泛化](https://wanghao.blog.csdn.net/article/details/124409374)

## EfficientNet系列

[【第34篇】 EfficientNetV2：更快、更小、更強(qiáng)——論文翻譯](https://wanghao.blog.csdn.net/article/details/117399085)

## MOBILEVIT

[【第35篇】MOBILEVIT：輕量、通用和適用移動設(shè)備的Vision Transformer](https://wanghao.blog.csdn.net/article/details/124546928)

## EdgeViTs

[【第37篇】EdgeViTs： 在移動設(shè)備上使用Vision Transformers 的輕量級 CNN](https://wanghao.blog.csdn.net/article/details/124730330)

## MixConv

[【第38篇】MixConv：混合深度卷積核](https://wanghao.blog.csdn.net/article/details/124779609)

## RepLKNet

[【第39篇】RepLKNet將內(nèi)核擴(kuò)展到 31x31：重新審視 CNN 中的大型內(nèi)核設(shè)計(jì)](https://wanghao.blog.csdn.net/article/details/124875771)

## TransFG

[【第40篇】TransFG：用于細(xì)粒度識別的 Transformer 架構(gòu)](https://wanghao.blog.csdn.net/article/details/124919932)

## ConvMAE

[【第41篇】ConvMAE：Masked Convolution 遇到 Masked Autoencoders](https://wanghao.blog.csdn.net/article/details/124988783)

## MicroNet

[【第42篇】MicroNet：以極低的 FLOP 實(shí)現(xiàn)圖像識別](https://wanghao.blog.csdn.net/article/details/125177445)

## RepVGG?

[【第46篇】RepVGG ：讓卷積再次偉大](https://wanghao.blog.csdn.net/article/details/126446922)

## MaxViT

[【第48篇】MaxViT：多軸視覺轉(zhuǎn)換器](https://wanghao.blog.csdn.net/article/details/127064117)

## MAFormer

[【第53篇】MAFormer: 基于多尺度注意融合的變壓器網(wǎng)絡(luò)視覺識別](https://wanghao.blog.csdn.net/article/details/127492341)

## GhostNet系列

[【第56篇】GhostNet:廉價(jià)操作得到更多的特征](https://wanghao.blog.csdn.net/article/details/127981705)

[【第57篇】RepGhost:一個(gè)通過重新參數(shù)化實(shí)現(xiàn)硬件高效的Ghost模塊](https://wanghao.blog.csdn.net/article/details/128090737)

## DEiT系列

[【第58篇】DEiT：通過注意力訓(xùn)練數(shù)據(jù)高效的圖像transformer &蒸餾](https://wanghao.blog.csdn.net/article/details/128180419)

## MetaFormer

[【第59篇】MetaFormer實(shí)際上是你所需要的視覺](https://wanghao.blog.csdn.net/article/details/128281326)

## RegNet

[【第60篇】RegNet：設(shè)計(jì)網(wǎng)絡(luò)設(shè)計(jì)空間](https://wanghao.blog.csdn.net/article/details/128339572)

# 注意力機(jī)制

[【第23篇】NAM：基于標(biāo)準(zhǔn)化的注意力模塊](https://wanghao.blog.csdn.net/article/details/122092352)

# 物體檢測

[【第6篇】SSD論文翻譯和代碼匯總](https://wanghao.blog.csdn.net/article/details/105788036)

[【第7篇】CenterNet](https://wanghao.blog.csdn.net/article/details/105593958)

[【第8篇】M2Det](https://wanghao.blog.csdn.net/article/details/105593927)

[【第9篇】YOLOX](https://wanghao.blog.csdn.net/article/details/119535667)

[【第11篇】微軟發(fā)布的Dynamic Head，創(chuàng)造COCO新記錄：60.6AP](https://wanghao.blog.csdn.net/article/details/120365468)

[【第12篇】Sparse R-CNN: End-to-End Object Detection with Learnable Proposals](https://wanghao.blog.csdn.net/article/details/120413137)

[【第13篇】CenterNet2論文解析，COCO成績最高56.4mAP](https://wanghao.blog.csdn.net/article/details/120464708)

[【第14篇】UMOP](https://wanghao.blog.csdn.net/article/details/120506747)

[【第15篇】CBNetV2](https://wanghao.blog.csdn.net/article/details/120583025)

[【第19篇 】SE-SSD論文翻譯](https://wanghao.blog.csdn.net/article/details/120875331)

[【第24篇】YOLOR：多任務(wù)的統(tǒng)一網(wǎng)絡(luò)](https://wanghao.blog.csdn.net/article/details/122357992)

[【第31篇】探索普通視覺Transformer Backbones用于物體檢測](https://wanghao.blog.csdn.net/article/details/123960815)

[【第36篇】CenterNet++ 用于對象檢測](https://wanghao.blog.csdn.net/article/details/124623781)

[【第45篇】YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors](https://wanghao.blog.csdn.net/article/details/126302859)

# 行人屬性識別

[【第66篇】行人屬性識別研究綜述（一）](https://wanghao.blog.csdn.net/article/details/128736760)

[【第66篇】行人屬性識別研究綜述（二）](https://wanghao.blog.csdn.net/article/details/128736732)

# 行人跟蹤

[【第47篇】BoT-SORT：強(qiáng)大的關(guān)聯(lián)多行人跟蹤](https://wanghao.blog.csdn.net/article/details/126890651)

[【第65篇】SMILEtrack:基于相似度學(xué)習(xí)的多目標(biāo)跟蹤](https://blog.csdn.net/hhhhhhhhhhwwwwwwwwww/article/details/128615947)

[【第70篇】DeepSort：論文翻譯](https://wanghao.blog.csdn.net/article/details/129003397)

# OCR

[【第20篇】像人類一樣閱讀：自主、雙向和迭代語言 場景文本識別建模](https://wanghao.blog.csdn.net/article/details/121313548)

[【第44篇】DBNet：具有可微分二值化的實(shí)時(shí)場景文本檢測](https://wanghao.blog.csdn.net/article/details/125513523)

# 超分辨采樣

[【第33篇】SwinIR](https://wanghao.blog.csdn.net/article/details/124434886)

# 弱光增強(qiáng)

## RetinexNet

[【第52篇】RetinexNet: Deep Retinex Decomposition for Low-Light Enhancement](https://wanghao.blog.csdn.net/article/details/127400091)

[【第50篇】邁向快速、靈活、穩(wěn)健的微光圖像增強(qiáng)](https://wanghao.blog.csdn.net/article/details/127211265)

# NLP

[【第17篇】TextCNN](https://wanghao.blog.csdn.net/article/details/120729088)

[【第18篇】Bert論文翻譯](https://wanghao.blog.csdn.net/article/details/120864338)

# 多模態(tài)

[【第43篇】CLIP：從自然語言監(jiān)督中學(xué)習(xí)可遷移的視覺模型](https://wanghao.blog.csdn.net/article/details/125452516)

# 知識蒸餾

[【第54篇】知識蒸餾：Distilling the Knowledge in a Neural Network](https://wanghao.blog.csdn.net/article/details/127808674)

# 剪枝

[【第55篇】剪枝算法：通過網(wǎng)絡(luò)瘦身學(xué)習(xí)高效卷積網(wǎng)絡(luò)](https://wanghao.blog.csdn.net/article/details/127871910)

# 智慧城市

[【第51篇】用于交通預(yù)測的時(shí)空交互動態(tài)圖卷積網(wǎng)絡(luò)](https://wanghao.blog.csdn.net/article/details/127306179)