• Content
• • ResNet Descriptions
  • Model Architecture
  • Dataset
  • Environment Requirements
  • Quick Start
  • Script Description
  • Script Parameters
  • Training Process
  • Evaluation Process
  • • Evaluation
  • Model Description
  • • Performance
    • • Training Performance
      • Inference Performance

Content

• ResNet Descriptions
• Model Architecture
• Dataset
• Environment Requirements
• Quick Start
• Script Description
• Script Parameters
• Training Process
• Evaluation Process
  • Evaluation
• Model Description
  • Performance
    • Training Performance
    • Inference Performance

ResNet Descriptions

ResNet was proposed in 2015, it is a typoe of neural network that directly consumes 2D images.

Paper Farooq M, Hafeez A. Covid-resnet: A deep learning framework for screening of covid19 from radiographs. arXiv preprint arXiv:2003.14395. 2020 Mar 31.

Model Architecture

The ResNet classification network takes n 2D images as input, applies input and feature transformations. BN is introdued before each ReLU.

Dataset

Dataset used: [Covid-19 CT Image Dataset]

• Dataset size: 1.2G
  • Train: 92M, 1716+1851 images and corresponding labels.
  • Test: 23M, 429+463
• Data format: png/jpg files
  • Notes: Data will be processed in src/dataset.py

Environment Requirements

• Hardware（Ascend）
  • Prepare hardware environment with Ascend processor.
• Framework
  • MindSpore
• For more information, please check the resources below：
  • MindSpore Tutorials
  • MindSpore Python API

• suwen package

  pip install -r requirements.txt
  pip install ./suwen-1.0.1-py3-none-any.whl
  

Quick Start

After installing MindSpore via the official website, you can start training and evaluation as follows:

# enter script dir, train PointNet
sh run_train_ascend.sh

# enter script dir, evaluate PointNet
sh run_eval.sh

Script Description

.
├── README.md
├── eval.py
├── eval_log.txt
├── resnet.ckpt
├── scripts
│   ├── run_eval.sh
│   └── run_train_ascend.sh
├── src
│   ├── __pycache__
│   │   ├── dataset.cpython-37.pyc
│   │   └── utils.cpython-37.pyc
│   ├── dataset.py
│   └── utils.py
├── suwen-1.0.1-py3-none-any.whl
├── train.py
└── train_log.txt

3 directories, 13 files

Script Parameters

Major parameters in train.py are as follows:

--data_path: The absolute full path to the train and evaluation datasets.
--ckpt_path: The absolute full path to the checkpoint file saved after training.

More hyperparamteters can be modified in src/config.py.

Training Process

• running on Ascend

  sh run_train_ascend.sh
  

  After training, the loss value will be achieved as what in train_log.txt

  The model checkpoint will be saved in the current ckpt directory.

Evaluation Process

Evaluation

Before running the command below, please check the checkpoint path used for evaluation.

• running on Ascend

  sh scripts/run_eval.sh
  

  You can view the results through the file "eval_log". The accuracy of the test dataset will be as what in eval_log.txt.

Model Description

Performance

Training Performance

Parameters
Resource	Ascend 910; CPU 2.60GHz, 24cores; Memory, 96G
uploaded Date	11/23/2021 (month/day/year)
MindSpore Version	1.3.0
Dataset	Covid-Classification
Training Parameters	epoch=100, batch_size =32 , lr=0.01
Optimizer	Adam
Loss Function	Softmax Cross Entropy
outputs	probability
Loss	SoftmaxCrossEntropyWithLogits
Speed	19ms/step
Total time	1900ms
Checkpoint for Fine tuning	180M (.ckpt file)

Inference Performance

Parameters
Resource	Ascend 910; CPU 2.60GHz, 24cores; Memory, 96G
uploaded Date	11/23/2021 (month/day/year)
MindSpore Version	1.3.0
Dataset	Covid-Classification
batch_size	1
outputs	probability
Accuracy	96.99%