Methods › Computer Vision › Image Model Blocks

Image Model Blocks

83 methods 4,655 papers tagged archive 2025-07-28

The archive attaches this collection's text per method and the copies differ: 3 distinct texts across 83 of the 83 methods here. All are shown, most-carried first (a tie goes to the text carrying Papers with Code's collection boilerplate, then to the longer text); no vote is taken between them.

Text 1, carried by 64 of 83 methods:

Image Model Blocks are building blocks used in image models such as convolutional neural networks. Below you can find a continuously updating list of image model blocks.

Text 2, carried by 18 of 83 methods:

Skip Connection Blocks are building blocks for neural networks that feature skip connections. These skip connections 'skip' some layers allowing gradients to better flow through the network. Below you will find a continuously updating list of skip connection blocks:

Text 3, carried by 1 of 83 methods:

Transformers are a type of neural network architecture that have several properties that make them effective for modeling data with long-range dependencies. They generally feature a combination of multi-headed attention mechanisms, residual connections, layer normalization, feedforward connections, and positional embeddings.

Methods

All 83 methods in this collection, most-tagged first. Year is the archive's introduced_year; the archive stores 2000 when it has none, shown here as “–”. Papers counts distinct papers the archive tags with the method. Click a heading to sort.


Residual Block	–	2,807
Bottleneck Residual Block	–	2,049
Squeeze-and-Excitation Block	–	543
Dense Block	–	497
Inception Module	–	206
Spatial Transformer	–	169
Non-Local Block	–	168
ResNeXt Block	–	132
PnP	–	114
Fire Module	–	95
Inception-v3 Module	–	95
Axial Attention	–	59
ShuffleNet Block	–	51
ESP Efficient Spatial Pyramid	–	46
CBAM Convolutional Block Attention Module	–	43
SRGAN Residual Block	–	34
Hourglass Module	–	31
Ghost Module	–	30
Reduction-A	–	28
Selective Kernel	–	27
Res2Net Block	–	25
ENet Bottleneck	–	24
ENet Dilated Bottleneck	–	24
ENet Initial Block	–	24
Ghost Bottleneck	–	24
ShuffleNet V2 Block	–	21
Split Attention	–	20
Inception-ResNet-v2 Reduction-B	–	18
Inception-ResNet-v2-A	–	18
Inception-ResNet-v2-B	–	18
Inception-ResNet-v2-C	–	18
ShuffleNet V2 Downsampling Block	–	14
Inception-A	–	13
Inception-B	–	13
Inception-C	–	13
Reduction-B	–	13
FBNet Block	–	12
Global Context Block	–	12
TNT Transformer in Transformer	–	12
One-Shot Aggregation	–	11
Bottleneck Transformer Block	–	10
Pyramidal Bottleneck Residual Unit	–	8
Pyramidal Residual Unit	–	8
Spatial Feature Transform	–	8
Mixer Layer MLP-Mixer Layer	–	7
Scale Aggregation Block	–	7
Fractal Block	–	6
Multiscale Dilated Convolution Block	–	5
NVAE Encoder Residual Cell	–	5
NVAE Generative Residual Cell	–	5
Big-Little Module	–	4
LRNet Local Relation Network	–	4
Local Patch Interaction	–	4
XCiT Layer	–	4
CSPResNeXt Block	–	3
EESP Extremely Efficient Spatial Pyramid of Depth-wise Dilated Separable Convolutions	–	3
SqueezeNeXt Block	–	3
Style-based Recalibration Module	–	3
Two-Way Dense Layer	–	3
DVD-GAN DBlock	–	2
DVD-GAN GBlock	–	2
Elastic Dense Block	–	2
Global Local Attention Module	–	2
AttLWB Attentional Liquid Warping Block	–	1
Compact Global Descriptor	–	1
Content-Conditioned Style Encoder	–	1
Contextual Residual Aggregation	–	1
CornerNet-Squeeze Hourglass Module	–	1
Depthwise Fire Module	–	1
DiCE Unit	–	1
DimFuse Dimension-wise Fusion	–	1
E-MBConv	–	1
Effective Squeeze-and-Excitation Block	–	1
Elastic ResNeXt Block	–	1
Harmonic Block	–	1
Hierarchical-Split Block	–	1
Mobile DenseNet	–	1
OSA (identity mapping + eSE)	–	1
Patch Merger Patch Merger Module	–	1
Residual SRM	–	1
Spatial Group-wise Enhance	–	1
Strided EESP	–	1
Efficient Channel Attention	–	0