Skip to content

README.md

Latest commit

 

History

History
199 lines (164 loc) · 5.32 KB

README.md

File metadata and controls

199 lines (164 loc) · 5.32 KB

An Test Tool for Module Analysis

Description：

  1. Calculate the number of parameters and floating point of the model, as well as the model inference time
  2. Visual feature map

installation

conda create -n analysis python=3.8 -y
pip install -r requirements.txt

torch==2.4.0 torchvision==0.19.0 torchaudio==2.4.0

Quickly Start

run compute_model_info.py

You can compute custom models or models officially implemented by pytorch

Details of the official pytorch model can be found in [Supported Models](#Supported models)

from utils.computing_function import computing_info

from model.conv import *
from model.block import *

if __name__ == '__main__':
    
    # Calculating custom model
    model = C3k2(c1=3, c2=64)  # replace your model/module
    computing_info(
        images_dir="./images",
        output_dir='./vis_result',
        model_name='C3k2',   # replace your model name
        device='cpu',
        model=model,
    )
    
    
	# Calculating the pytorch official implementation model
    computing_info(
        images_dir="./images",
        output_dir='./vis_result',
        model_name='resnet50',  
        device='cpu',
        model=model,
    )

Case presentation

visualization of some famous backbone

1

visualization of some famous block/Conv operation

2

Supported models

more details can be found in model/model_zoo.py

if model_name == 'resnet50':
    model = models.resnet50()
elif model_name == 'resnet101':
    model = models.resnet101()
elif model_name == 'vgg16':
    model = models.vgg16()
elif model_name == 'densenet121':
    model = models.densenet121()
elif model_name == 'mobilenet_v2':
    model = models.mobilenet_v2()
elif model_name == 'resnet18':
    model = models.resnet18()
elif model_name == 'resnet34':
    model = models.resnet34()
elif model_name == 'resnet152':
    model = models.resnet152()
elif model_name == 'vgg11':
    model = models.vgg11()
elif model_name == 'vgg13':
    model = models.vgg13()
elif model_name == 'vgg19':
    model = models.vgg19()
elif model_name == 'densenet169':
    model = models.densenet169()
elif model_name == 'densenet201':
    model = models.densenet201()
elif model_name == 'densenet161':
    model = models.densenet161()
elif model_name == 'mobilenet_v3_large':
    model = models.mobilenet_v3_large()
elif model_name == 'mobilenet_v3_small':
    model = models.mobilenet_v3_small()
elif model_name == 'inception_v3':
    model = models.inception_v3()
elif model_name == 'efficientnet_b0':
    model = models.efficientnet_b0()
elif model_name == 'efficientnet_b1':
    model = models.efficientnet_b1()
elif model_name == 'efficientnet_b7':
    model = models.efficientnet_b7()
elif model_name == 'shufflenet_v2_x0_5':
    model = models.shufflenet_v2_x0_5()
elif model_name == 'shufflenet_v2_x1_0':
    model = models.shufflenet_v2_x1_0()
elif model_name == 'squeezenet1_0':
    model = models.squeezenet1_0()
elif model_name == 'squeezenet1_1':
    model = models.squeezenet1_1()
elif model_name == 'alexnet':
    model = models.alexnet()
elif model_name == 'googlenet':
    model = models.googlenet()
elif model_name == 'regnet_y_400mf':
    model = models.regnet_y_400mf()
elif model_name == 'regnet_y_800mf':
    model = models.regnet_y_800mf()
elif model_name == 'regnet_x_1_6gf':
    model = models.regnet_x_1_6gf()
elif model_name == 'convnext_tiny':
    model = models.convnext_tiny()
elif model_name == 'convnext_small':
    model = models.convnext_small()
elif model_name == 'convnext_base':
    model = models.convnext_base()
elif model_name == 'convnext_large':
    model = models.convnext_large()
elif model_name == 'efficientnet_v2_s':
    model = models.efficientnet_v2_s()
elif model_name == 'efficientnet_v2_m':
    model = models.efficientnet_v2_m()
elif model_name == 'efficientnet_v2_l':
    model = models.efficientnet_v2_l()
elif model_name == 'mnasnet0_5':
    model = models.mnasnet0_5()
elif model_name == 'mnasnet0_75':
    model = models.mnasnet0_75()
elif model_name == 'mnasnet1_0':
    model = models.mnasnet1_0()
elif model_name == 'mnasnet1_3':
    model = models.mnasnet1_3()
elif model_name == 'regnet_y_1_6gf':
    model = models.regnet_y_1_6gf()
elif model_name == 'regnet_y_3_2gf':
    model = models.regnet_y_3_2gf()
elif model_name == 'regnet_y_16gf':
    model = models.regnet_y_16gf()
elif model_name == 'regnet_x_3_2gf':
    model = models.regnet_x_3_2gf()
elif model_name == 'regnet_x_8gf':
    model = models.regnet_x_8gf()
elif model_name == 'vit_b_16':
    model = models.vit_b_16()
elif model_name == 'vit_b_32':
    model = models.vit_b_32()
elif model_name == 'vit_l_16':
    model = models.vit_l_16()
elif model_name == 'vit_l_32':
    model = models.vit_l_32()
elif model_name == 'swin_t':
    model = models.swin_t()
elif model_name == 'swin_s':
    model = models.swin_s()
elif model_name == 'swin_b':
    model = models.swin_b()
elif model_name == 'swin_l':
    model = models.swin_l()
else:
    print(f"Model {model_name} is not in model zoo.")

Contact

If you have any question or suggestion, please feel free to contact us by raising an issue or sending an email to glory947446@gmail.com.

License

This res is licensed under the MIT License. See the LICENSE file for more details.