Skip to content

Model_Architecture.md

Latest commit

 

History

History
78 lines (47 loc) · 5.08 KB

Model_Architecture.md

File metadata and controls

78 lines (47 loc) · 5.08 KB
graph LR
    DetNAS_Model_Series["DetNAS Model Series"]
    OneShot_Model_Series["OneShot Model Series"]
    ShuffleNetV1_Model_Series["ShuffleNetV1 Model Series"]
    ShuffleNetV2_Model_Series["ShuffleNetV2 Model Series"]
    Shared_Network_Utilities["Shared Network Utilities"]
    DetNAS_Model_Series -- "integrates and utilizes" --> Shared_Network_Utilities
    OneShot_Model_Series -- "integrates and utilizes" --> Shared_Network_Utilities
    ShuffleNetV1_Model_Series -- "integrates and utilizes" --> Shared_Network_Utilities
    ShuffleNetV2_Model_Series -- "integrates and utilizes" --> Shared_Network_Utilities
    Shared_Network_Utilities -- "is utilized by" --> DetNAS_Model_Series
    Shared_Network_Utilities -- "is utilized by" --> OneShot_Model_Series
    Shared_Network_Utilities -- "is utilized by" --> ShuffleNetV1_Model_Series
    Shared_Network_Utilities -- "is utilized by" --> ShuffleNetV2_Model_Series
Loading

CodeBoardingDemoContact

Details

The ShuffleNet-Series project is structured around distinct neural network model architectures, each encapsulated within its own dedicated component. These include DetNAS Model Series, OneShot Model Series, ShuffleNetV1 Model Series, and ShuffleNetV2 Model Series (including its Large and ExLarge variants). Each model series defines its unique network structure and specialized building blocks. A Shared Network Utilities component provides fundamental, reusable operations, such as channel shuffling, that are common across these various ShuffleNet-based architectures, facilitating efficient and consistent implementation of core functionalities.

DetNAS Model Series

Encapsulates the complete neural network architecture for the DetNAS model, defining its overall structure and specialized building blocks.

Related Classes/Methods:

OneShot Model Series

Encapsulates the complete neural network architecture for the OneShot model, defining its overall structure and specialized building blocks.

Related Classes/Methods:

ShuffleNetV1 Model Series

Encapsulates the complete neural network architecture for the ShuffleNetV1 model, defining its overall structure and specialized building blocks.

Related Classes/Methods:

ShuffleNetV2 Model Series

Defines the comprehensive neural network architecture for the ShuffleNetV2 model, including its base structure and specific variants (Large, ExLarge), along with its core building blocks.

Related Classes/Methods:

Shared Network Utilities

Provides fundamental, reusable operations, specifically channel shuffling, that are common across various ShuffleNet-based architectures.

Related Classes/Methods:

FAQ