Very Deep Convolutional Networks for Large-Scale Image Recognition

Led to

• Captions to Visual Concepts2014 · cited 5×
  “Next, we featurize each of these regions using rich convolutional neural network (CNN) features, fine-tuned on our training data krizhevskyNIPS12; simonyan14very.”

  From Captions to Visual Concepts · §Introduction
• Karpathy visual-semantic alignment2014 · cited 3×
  “We list their performance with a CNN that is equivalent in power (AlexNet krizhevsky2012imagenet) to the one used in this work, though similar to vinyals2014show they outperform our model with a more powerful CNN (VGGNet…”

  From Karpathy visual-semantic alignment · §Experiments
• PReLU / He init2015 · cited 21×, 14 in Method
  “With fixed standard deviations (e.g., 0.01 in Krizhevsky2012), very deep models (e.g., >>8 conv layers) have difficulties to converge, as reported by the VGG team Simonyan2014 and also observed in our experiments.”

  From PReLU / He init · §Approach
• Learning like a Child2015 · cited 2×, 1 in Method
  “The vision component contains a 16-layer deep convolutional neural network (CNN [48]) pre-trained on the ImageNet classification task [45].”

  From Learning like a Child · §The Image Captioning Model
• VQA2015 · cited 3×, 3 in Method
  “I: The activations from the last hidden layer of VGGNet [48] are used as 4096-dim image embedding.”

  From VQA · §V VQA Baselines and Methods
• Flickr30k Entities2015 · cited 2×
  “By using better region features (Fast RCNN (Girshick, 2015) instead of ImageNet-trained VGG (Simonyan and Zisserman, 2014)) in combination with size and color cues, we are able to improve the Recall@1 for phrase localiza…”

  From Flickr30k Entities · §Introduction
• FM-IQA2015 · cited 2×, 1 in Method
  “For computer vision, methods based on Convolutional Neural Network (CNN [20]) achieve the state-of-the-art performance in various tasks, such as object classification [17, 34, 17], detection [10, 44] and segmentation [3]…”

  From FM-IQA · §Related Work
• Inception v32015 · cited 2×
  “VGGNet simonyan2014very and GoogLeNet szegedy2015going yielded similarly high performance in the 2014 ILSVRC russakovsky2014imagenet classification challenge.”

  From Inception v3 · §Introduction
• ResNet2015 · cited 11×
  “Our plain baselines (Fig. 3, middle) are mainly inspired by the philosophy of VGG nets Simonyan2015 (Fig. 3, left).”

  From ResNet · §Deep Residual Learning
• Visual Genome2016 · cited 5×
  “Much progress has been made in recent years towards this goal, including image classification Deng et al., 2009; Perronnin et al., 2010; Simonyan and Zisserman, 2014; Krizhevsky et al., 2012; Szegedy et al., 2014 and obj…”

  From Visual Genome · §Introduction
• Wide ResNet2016 · cited 4×
  “Convolutional neural networks have seen a gradual increase of the number of layers in the last few years, starting from AlexNet [Krizhevsky et al.(2012a)Krizhevsky, Sutskever, and Hinton], VGG [Simonyan and Zisserman(201…”

  From Wide ResNet · §Introduction
• ResNeXt2016 · cited 2×
  “The VGG-nets Simonyan2015 exhibit a simple yet effective strategy of constructing very deep networks: stacking building blocks of the same shape.”

  From ResNeXt · §Introduction
• Constrained beam search captioning2016 · cited 1×, 1 in Method
  “For the CNN component of the model, we evaluate using the 16-layer VGG Simonyan and Zisserman 2015 model and the 50-layer Residual Net He et al. 2016, pretrained on ILSVRC-2012 Russakovsky et al. 2015 in both cases.”

  From Constrained beam search captioning · §Approach
• VQA v22016 · cited 3×, 1 in Method
  “It should be noted that MCB uses image features from a more powerful CNN architecture ResNet ResNet while the previous two models use image features from VGGNet Simonyan15.”

  From VQA v2 · §Benchmarking Existing VQA Models
• MobileNets2017 · cited 2×
  “The general trend has been to make deeper and more complicated networks in order to achieve higher accuracy simonyan2014very; szegedy2015rethinking; szegedy2016inception; he2015deep.”

  From MobileNets · §Introduction
• Goyal large-batch SGD2017 · cited 2×
  “We are in an unprecedented era in AI research history in which the increasing data and model scale is rapidly improving accuracy in computer vision Krizhevsky2012; Zeiler2014; Sermanet2014; Simonyan2015; Szegedy2015; He2…”

  From Goyal large-batch SGD · §Introduction
• NASNet2017 · cited 4×, 1 in Method
  “Starting from the seminal work of krizhevsky2012imagenet on using convolutional architectures fukushima1982neocognitron; lecun1998gradient for ImageNet deng2009imagenet classification, successive advancements through arc…”

  From NASNet · §Introduction
• MobileNetV22018 · cited 2×
  “Both manual architecture search and improvements in training algorithms, carried out by numerous teams has lead to dramatic improvements over early designs such as AlexNet AlexNet, VGGNet VGGNet, GoogLeNet GoogleNet. , a…”

  From MobileNetV2 · §Related Work
• Instance discrimination2018 · cited 2×
  “As the network architecture has a big impact on the performance, we consider a few typical architectures: AlexNet krizhevsky2012imagenet, VGG16 simonyan2014very, ResNet-18, and ResNet-50 he2015deep.”

  From Instance discrimination · §Experiments
• ImageNet-trained CNNs are biased towards2018 · cited 1×, 1 in Method
  “The same images were fed to four CNNs pre-trained on standard ImageNet, namely AlexNet (Krizhevsky et al. 2012), GoogLeNet (Szegedy et al. 2015), VGG-16 (Simonyan & Zisserman 2015) and ResNet-50 (He et al. 2015).”

  From ImageNet-trained CNNs are biased towards · §Methods
• ImageNetV22019 · cited 2×
  “The models include the seminal AlexNet [36], widely used convolutional networks [49, 21, 27, 52], and the state-of-the-art [8, 39].”

  From ImageNetV2 · §Summary of Our Experiments
• Grid features for VQA2020 · cited 2×
  “As a result, images are represented by a collection of bounding box or region11 1 We use the terms ‘region’ and ‘bounding box’ interchangeably.-based features anderson2018bottom; teney2018tips–in contrast to vanilla grid…”

  From Grid features for VQA · §Introduction
• BYOL2020 · cited 2×
  “[1, 2, 3, 4, 5, 6, 7]”

  From BYOL · §Introduction
• Natural distribution shift robustness2020 · cited 1×, 1 in Method
  “This category includes 78 models with architectures ranging from AlexNet to EfficietNet, e.g., [50, 78, 37, 85, 88].”

  From Natural distribution shift robustness · §Experimental setup
• Swin2021 · cited 2×, 1 in Method
  “These stages jointly produce a hierarchical representation, with the same feature map resolutions as those of typical convolutional networks, e.g., VGG [52] and ResNet [30].”

  From Swin · §Method
• Swin V22021 · cited 2×
  “While it has long been recognized that larger vision models usually perform better on vision tasks simonyan2014vgg; he2015resnet, the absolute model size was just able to reach about 1-2 billion parameters very recently…”

  From Swin V2 · §Introduction