mindspore/model_zoo/official/cv/mobilenetv1

Mobilenet_V1

• Mobilenet_V1
  • MobileNetV1 Description
  • Model architecture
  • Dataset
  • [Features]
    • [Mixed Precision(Ascend)]
  • [Environment Requirements]
  • [Script description]
    • [Script and sample code]
  • Training process
    • Usage
    • Launch
    • Result
  • Evaluation process
    • Usage
    • Launch
    • Result
  • [Model description]
    • Performance
      • Training Performance
  • Description of Random Situation
  • ModelZoo Homepage

MobileNetV1 Description

MobileNetV1 is a efficient network for mobile and embedded vision applications. MobileNetV1 is based on a streamlined architecture that uses depth-wise separable convolutions to build light weight deep n.eural networks

Paper Howard A G , Zhu M , Chen B , et al. MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications[J]. 2017.

Model architecture

The overall network architecture of MobileNetV1 is show below:

Link

Dataset

Dataset used: ImageNet2012

• Dataset size 224*224 colorful images in 1000 classes
  • Train：1,281,167 images
  • Test： 50,000 images
• Data format：jpeg
  • Note：Data will be processed in dataset.py
• Download the dataset, the directory structure is as follows:

└─dataset
    ├─ilsvrc                # train dataset
    └─validation_preprocess # evaluate dataset

[Features]

[Mixed Precision(Ascend)]

The mixed precision training method accelerates the deep learning neural network training process by using both the single-precision and half-precision data formats, and maintains the network precision achieved by the single-precision training at the same time. Mixed precision training can accelerate the computation process, reduce memory usage, and enable a larger model or batch size to be trained on specific hardware. For FP16 operators, if the input data type is FP32, the backend of MindSpore will automatically handle it with reduced precision. Users could check the reduced-precision operators by enabling INFO log and then searching ‘reduce precision’.

[Environment Requirements]

• Hardware（Ascend）
  • Prepare hardware environment with Ascend. If you want to try Ascend, please send the application form to ascend@huawei.com. Once approved, you can get the resources.
• Framework
  • MindSpore
• For more information, please check the resources below：
  • MindSpore Tutorials
  • MindSpore Python API

[Script description]

[Script and sample code]

├── MobileNetV1
  ├── README.md              # descriptions about MobileNetV1
  ├── scripts
  │   ├──run_distribute_train.sh        # shell script for distribute train
  │   ├──run_standalone_train.sh        # shell script for standalone train
  │   ├──run_eval.sh                    # shell script for evaluation
  ├── src
  │   ├──config.py           # parameter configuration
  │   ├──dataset.py          # creating dataset
  │   ├──lr_generator.py     # learning rate config
  │   ├──mobilenet_v1_fpn.py      # MobileNetV1 architecture
  │   ├──CrossEntropySmooth.py           # loss function
  ├── train.py               # training script
  ├── eval.py                # evaluation script

Training process

Usage

You can start training using python or shell scripts. The usage of shell scripts as follows:

• Ascend: sh run_distribute_train.sh [cifar10|imagenet2012] [RANK_TABLE_FILE] [DATASET_PATH] [PRETRAINED_CKPT_PATH] (optional)

For distributed training, a hccl configuration file with JSON format needs to be created in advance.

Please follow the instructions in the link hccn_tools.

Launch

# training example
  python:
      Ascend: python train.py --platform Ascend --dataset_path [TRAIN_DATASET_PATH]

  shell:
     Ascend: sh run_distribute_train.sh [cifar10|imagenet2012] [RANK_TABLE_FILE] [DATASET_PATH] [PRETRAINED_CKPT_PATH](optional)

Result

Training result will be stored in the example path. Checkpoints will be stored at ckpt_* by default, and training log will be wrote to ./train_parallel*/log with the platform Ascend .

epoch: 89 step: 1251, loss is 2.1829057
Epoch time: 146826.802, per step time: 117.368
epoch: 90 step: 1251, loss is 2.3499017
Epoch time: 150950.623, per step time: 120.664

Evaluation process

Usage

You can start training using python or shell scripts.If the train method is train or fine tune, should not input the [CHECKPOINT_PATH] The usage of shell scripts as follows:

• Ascend: sh run_eval.sh [cifar10|imagenet2012] [DATASET_PATH] [CHECKPOINT_PATH]

Launch

# eval example
  python:
      Ascend: python eval.py --dataset [cifar10|imagenet2012] --dataset_path [VAL_DATASET_PATH] --pretrain_ckpt [CHECKPOINT_PATH]

  shell:
      Ascend: sh run_eval.sh [cifar10|imagenet2012] [DATASET_PATH] [CHECKPOINT_PATH]

checkpoint can be produced in training process.

Result

Inference result will be stored in the example path, you can find result like the followings in eval/log.

result: {'top_5_accuracy': 0.9010016025641026, 'top_1_accuracy': 0.7128004807692307} ckpt=./train_parallel0/ckpt_0/mobilenetv1-90_1251.ckpt

[Model description]

Performance

Training Performance

Parameters	MobilenetV1
Model Version	V1
Resource	Ascend 910 * 4, cpu:2.60GHz 192cores, memory:755G
uploaded Date	11/28/2020
MindSpore Version	1.0.0
Dataset	ImageNet2012
Training Parameters	src/config.py
Optimizer	Momentum
Loss Function	SoftmaxCrossEntropy
outputs	probability
Loss	2.3499017
Accuracy	ACC1[71.28%]
Total time	225 min
Params (M)	3.3 M
Checkpoint for Fine tuning	27.3 M
Scripts	Link

Description of Random Situation

In train.py, we set the seed which is used by numpy.random, mindspore.common.Initializer, mindspore.ops.composite.random_ops and mindspore.nn.probability.distribution.

ModelZoo Homepage

Please check the official homepage.