SqueezeNet

Introduced by Iandola et al. in SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MB model size

SqueezeNet is a convolutional neural network that employs design strategies to reduce the number of parameters, notably with the use of fire modules that "squeeze" parameters using 1x1 convolutions.

Source: SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MB model size

Read Paper See Code

Papers

Paper	Code	Results	Date	Stars

Tasks

Task	Papers	Share
General Classification	11	10.58%
Object Detection	10	9.62%
Classification	8	7.69%
Image Classification	7	6.73%
Quantization	4	3.85%
Face Recognition	3	2.88%
Face Verification	3	2.88%
Specificity	3	2.88%
Object Recognition	3	2.88%

Usage Over Time

This feature is experimental; we are continuously improving our matching algorithm.

Components

Component	Type	Add Remove
1x1 Convolution	Convolutions
Convolution	Convolutions
Dropout	Regularization
Fire Module	Image Model Blocks
Global Average Pooling	Pooling Operations
Max Pooling	Pooling Operations
Residual Connection	Skip Connections
Softmax	Output Functions
Xavier Initialization	Initialization

Categories

Add Remove

Convolutional Neural Networks

Light-weight neural networks