From 37f78ec3e74ce4ad2ad1f7f66ea9f864f8bc46e4 Mon Sep 17 00:00:00 2001 From: zhaoting Date: Wed, 19 Aug 2020 17:30:20 +0800 Subject: [PATCH] mobilenetv2 mobilenetv3 readme normalize, delete mobilenetv3 ascend --- model_zoo/official/cv/mobilenetv2/Readme.md | 99 ++++++++---- model_zoo/official/cv/mobilenetv2/eval.py | 2 +- .../cv/mobilenetv2/scripts/run_infer.sh | 2 +- .../official/cv/mobilenetv2/src/dataset.py | 4 +- model_zoo/official/cv/mobilenetv2/train.py | 7 +- model_zoo/official/cv/mobilenetv3/Readme.md | 148 +++++++++++------- model_zoo/official/cv/mobilenetv3/eval.py | 19 +-- model_zoo/official/cv/mobilenetv3/train.py | 16 +- 8 files changed, 174 insertions(+), 123 deletions(-) diff --git a/model_zoo/official/cv/mobilenetv2/Readme.md b/model_zoo/official/cv/mobilenetv2/Readme.md index b39013b07e..f7021170fd 100644 --- a/model_zoo/official/cv/mobilenetv2/Readme.md +++ b/model_zoo/official/cv/mobilenetv2/Readme.md @@ -1,17 +1,37 @@ -# MobileNetV2 Description +# Contents + +- [MobileNetV2 Description](#mobilenetv2-description) +- [Model Architecture](#model-architecture) +- [Dataset](#dataset) +- [Features](#features) + - [Mixed Precision](#mixed-precision) +- [Environment Requirements](#environment-requirements) +- [Script Description](#script-description) + - [Script and Sample Code](#script-and-sample-code) + - [Training Process](#training-process) + - [Evaluation Process](#evaluation-process) + - [Evaluation](#evaluation) +- [Model Description](#model-description) + - [Performance](#performance) + - [Training Performance](#evaluation-performance) + - [Inference Performance](#evaluation-performance) +- [Description of Random Situation](#description-of-random-situation) +- [ModelZoo Homepage](#modelzoo-homepage) + +# [MobileNetV2 Description](#contents) MobileNetV2 is tuned to mobile phone CPUs through a combination of hardware- aware network architecture search (NAS) complemented by the NetAdapt algorithm and then subsequently improved through novel architecture advances.Nov 20, 2019. [Paper](https://arxiv.org/pdf/1905.02244) Howard, Andrew, Mark Sandler, Grace Chu, Liang-Chieh Chen, Bo Chen, Mingxing Tan, Weijun Wang et al. "Searching for MobileNetV2." In Proceedings of the IEEE International Conference on Computer Vision, pp. 1314-1324. 2019. -# Model architecture +# [Model architecture](#contents) The overall network architecture of MobileNetV2 is show below: [Link](https://arxiv.org/pdf/1905.02244) -# Dataset +# [Dataset](#contents) Dataset used: [imagenet](http://www.image-net.org/) @@ -22,10 +42,14 @@ Dataset used: [imagenet](http://www.image-net.org/) - Note: Data will be processed in src/dataset.py -# Features +# [Features](#contents) +## [Mixed Precision(Ascend)](#contents) -# Environment Requirements +The [mixed precision](https://www.mindspore.cn/tutorial/zh-CN/master/advanced_use/mixed_precision.html) 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](#contents) - Hardware（Ascend/GPU） - Prepare hardware environment with Ascend or GPU processor. If you want to try Ascend , please send the [application form](https://obs-9be7.obs.cn-east-2.myhuaweicloud.com/file/other/Ascend%20Model%20Zoo%E4%BD%93%E9%AA%8C%E8%B5%84%E6%BA%90%E7%94%B3%E8%AF%B7%E8%A1%A8.docx) to ascend@huawei.com. Once approved, you can get the resources. @@ -36,30 +60,33 @@ Dataset used: [imagenet](http://www.image-net.org/) - [MindSpore API](https://www.mindspore.cn/api/zh-CN/master/index.html) -# Script description +# [Script description](#contents) -## Script and sample code +## [Script and sample code](#contents) ```python ├── MobileNetV2 - ├── Readme.md + ├── Readme.md # descriptions about MobileNetV2 ├── scripts - │ ├──run_train.sh - │ ├──run_eval.sh + │ ├──run_train.sh # shell script for train + │ ├──run_eval.sh # shell script for evaluation ├── src - │ ├──config.py - │ ├──dataset.py - │ ├──luanch.py - │ ├──lr_generator.py - │ ├──mobilenetV2.py - ├── train.py - ├── eval.py + │ ├──config.py # parameter configuration + │ ├──dataset.py # creating dataset + │ ├──launch.py # start python script + │ ├──lr_generator.py # learning rate config + │ ├──mobilenetV2.py # MobileNetV2 architecture + ├── train.py # training script + ├── eval.py # evaluation script ``` -## Training process +## [Training process](#contents) ### Usage + +You can start training using python or shell scripts. The usage of shell scripts as follows: + - Ascend: sh run_train.sh Ascend [DEVICE_NUM] [VISIABLE_DEVICES(0,1,2,3,4,5,6,7)] [RANK_TABLE_FILE] [DATASET_PATH] [CKPT_PATH] - GPU: sh run_trian.sh GPU [DEVICE_NUM] [VISIABLE_DEVICES(0,1,2,3,4,5,6,7)] [DATASET_PATH] @@ -67,8 +94,13 @@ Dataset used: [imagenet](http://www.image-net.org/) ``` # training example - Ascend: sh run_train.sh Ascend 8 0,1,2,3,4,5,6,7 hccl_config.json ~/imagenet/train/ mobilenet_199.ckpt - GPU: sh run_train.sh GPU 8 0,1,2,3,4,5,6,7 ~/imagenet/train/ + python: + Ascend: python train.py --dataset_path ~/imagenet/train/ --device_targe Ascend + GPU: python train.py --dataset_path ~/imagenet/train/ --device_targe GPU + + shell: + Ascend: sh run_train.sh Ascend 8 0,1,2,3,4,5,6,7 hccl_config.json ~/imagenet/train/ mobilenet_199.ckpt + GPU: sh run_train.sh GPU 8 0,1,2,3,4,5,6,7 ~/imagenet/train/ ``` ### Result @@ -82,10 +114,12 @@ epoch: [ 1/200], step:[ 624/ 625], loss:[3.917/3.917], time:[138221.250], lr: epoch time: 138331.250, per step time: 221.330, avg loss: 3.917 ``` -## Eval process +## [Eval process](#contents) ### Usage +You can start training using python or shell scripts. The usage of shell scripts as follows: + - Ascend: sh run_infer.sh Ascend [DATASET_PATH] [CHECKPOINT_PATH] - GPU: sh run_infer.sh GPU [DATASET_PATH] [CHECKPOINT_PATH] @@ -93,8 +127,13 @@ epoch time: 138331.250, per step time: 221.330, avg loss: 3.917 ``` # infer example - Ascend: sh run_infer.sh Ascend ~/imagenet/val/ ~/train/mobilenet-200_625.ckpt - GPU: sh run_infer.sh GPU ~/imagenet/val/ ~/train/mobilenet-200_625.ckpt + python: + Ascend: python eval.py --dataset_path ~/imagenet/val/ --checkpoint_path mobilenet_199.ckpt --device_targe Ascend + GPU: python eval.py --dataset_path ~/imagenet/val/ --checkpoint_path mobilenet_199.ckpt --device_targe GPU + + shell: + Ascend: sh run_infer.sh Ascend ~/imagenet/val/ ~/train/mobilenet-200_625.ckpt + GPU: sh run_infer.sh GPU ~/imagenet/val/ ~/train/mobilenet-200_625.ckpt ``` > checkpoint can be produced in training process. @@ -107,9 +146,9 @@ Inference result will be stored in the example path, you can find result like th result: {'acc': 0.71976314102564111} ckpt=/path/to/checkpoint/mobilenet-200_625.ckpt ``` -# Model description +# [Model description](#contents) -## Performance +## [Performance](#contents) ### Training Performance @@ -147,5 +186,11 @@ result: {'acc': 0.71976314102564111} ckpt=/path/to/checkpoint/mobilenet-200_625. | Total time | | | | | Model for inference | | | | -# ModelZoo Homepage - [Link](https://gitee.com/mindspore/mindspore/tree/master/mindspore/model_zoo) \ No newline at end of file +# [Description of Random Situation](#contents) + +In dataset.py, we set the seed inside “create_dataset" function. We also use random seed in train.py. + +# [ModelZoo Homepage](#contents) + +Please check the official [homepage](https://gitee.com/mindspore/mindspore/tree/master/model_zoo). + \ No newline at end of file diff --git a/model_zoo/official/cv/mobilenetv2/eval.py b/model_zoo/official/cv/mobilenetv2/eval.py index 6431f6bfd3..897e7ffe27 100644 --- a/model_zoo/official/cv/mobilenetv2/eval.py +++ b/model_zoo/official/cv/mobilenetv2/eval.py @@ -39,7 +39,7 @@ if __name__ == '__main__': net = None if args_opt.device_target == "Ascend": config = config_ascend - device_id = int(os.getenv('DEVICE_ID')) + device_id = int(os.getenv('DEVICE_ID', '0')) context.set_context(mode=context.GRAPH_MODE, device_target="Ascend", device_id=device_id, save_graphs=False) net = mobilenet_v2(num_classes=config.num_classes, device_target="Ascend") diff --git a/model_zoo/official/cv/mobilenetv2/scripts/run_infer.sh b/model_zoo/official/cv/mobilenetv2/scripts/run_infer.sh index e13aa981e6..c596e33a23 100644 --- a/model_zoo/official/cv/mobilenetv2/scripts/run_infer.sh +++ b/model_zoo/official/cv/mobilenetv2/scripts/run_infer.sh @@ -47,7 +47,7 @@ fi mkdir ../eval cd ../eval || exit -# luanch +# launch python ${BASEPATH}/../eval.py \ --device_target=$1 \ --dataset_path=$2 \ diff --git a/model_zoo/official/cv/mobilenetv2/src/dataset.py b/model_zoo/official/cv/mobilenetv2/src/dataset.py index 887d86a5e1..528ead053c 100644 --- a/model_zoo/official/cv/mobilenetv2/src/dataset.py +++ b/model_zoo/official/cv/mobilenetv2/src/dataset.py @@ -35,8 +35,8 @@ def create_dataset(dataset_path, do_train, config, device_target, repeat_num=1, dataset """ if device_target == "Ascend": - rank_size = int(os.getenv("RANK_SIZE")) - rank_id = int(os.getenv("RANK_ID")) + rank_size = int(os.getenv("RANK_SIZE", '1')) + rank_id = int(os.getenv("RANK_ID", '0')) if rank_size == 1: ds = de.ImageFolderDatasetV2(dataset_path, num_parallel_workers=8, shuffle=True) else: diff --git a/model_zoo/official/cv/mobilenetv2/train.py b/model_zoo/official/cv/mobilenetv2/train.py index 5ea5e6d2c3..4fb800d6dd 100644 --- a/model_zoo/official/cv/mobilenetv2/train.py +++ b/model_zoo/official/cv/mobilenetv2/train.py @@ -53,11 +53,10 @@ parser.add_argument('--device_target', type=str, default=None, help='run device_ args_opt = parser.parse_args() if args_opt.device_target == "Ascend": - device_id = int(os.getenv('DEVICE_ID')) - rank_id = int(os.getenv('RANK_ID')) - rank_size = int(os.getenv('RANK_SIZE')) + device_id = int(os.getenv('DEVICE_ID', '0')) + rank_id = int(os.getenv('RANK_ID', '0')) + rank_size = int(os.getenv('RANK_SIZE', '1')) run_distribute = rank_size > 1 - device_id = int(os.getenv('DEVICE_ID')) context.set_context(mode=context.GRAPH_MODE, device_target="Ascend", device_id=device_id, save_graphs=False) diff --git a/model_zoo/official/cv/mobilenetv3/Readme.md b/model_zoo/official/cv/mobilenetv3/Readme.md index 7031faa9e4..7c82402d67 100644 --- a/model_zoo/official/cv/mobilenetv3/Readme.md +++ b/model_zoo/official/cv/mobilenetv3/Readme.md @@ -1,17 +1,35 @@ -# MobileNetV3 Description +# Contents + +- [MobileNetV3 Description](#mobilenetv3-description) +- [Model Architecture](#model-architecture) +- [Dataset](#dataset) +- [Environment Requirements](#environment-requirements) +- [Script Description](#script-description) + - [Script and Sample Code](#script-and-sample-code) + - [Training Process](#training-process) + - [Evaluation Process](#evaluation-process) + - [Evaluation](#evaluation) +- [Model Description](#model-description) + - [Performance](#performance) + - [Training Performance](#evaluation-performance) + - [Inference Performance](#evaluation-performance) +- [Description of Random Situation](#description-of-random-situation) +- [ModelZoo Homepage](#modelzoo-homepage) + +# [MobileNetV3 Description](#contents) MobileNetV3 is tuned to mobile phone CPUs through a combination of hardware- aware network architecture search (NAS) complemented by the NetAdapt algorithm and then subsequently improved through novel architecture advances.Nov 20, 2019. [Paper](https://arxiv.org/pdf/1905.02244) Howard, Andrew, Mark Sandler, Grace Chu, Liang-Chieh Chen, Bo Chen, Mingxing Tan, Weijun Wang et al. "Searching for mobilenetv3." In Proceedings of the IEEE International Conference on Computer Vision, pp. 1314-1324. 2019. -# Model architecture +# [Model architecture](#contents) The overall network architecture of MobileNetV3 is show below: [Link](https://arxiv.org/pdf/1905.02244) -# Dataset +# [Dataset](#contents) Dataset used: [imagenet](http://www.image-net.org/) @@ -22,10 +40,7 @@ Dataset used: [imagenet](http://www.image-net.org/) - Note: Data will be processed in src/dataset.py -# Features - - -# Environment Requirements +# [Environment Requirements](#contents) - Hardware（GPU） - Prepare hardware environment with GPU processor. @@ -36,37 +51,42 @@ Dataset used: [imagenet](http://www.image-net.org/) - [MindSpore API](https://www.mindspore.cn/api/zh-CN/master/index.html) -# Script description +# [Script description](#contents) -## Script and sample code +## [Script and sample code](#contents) ```python -├── MobilenetV3 - ├── Readme.md +├── MobileNetV3 + ├── Readme.md # descriptions about MobileNetV3 ├── scripts - │ ├──run_train.sh - │ ├──run_eval.sh + │ ├──run_train.sh # shell script for train + │ ├──run_eval.sh # shell script for evaluation ├── src - │ ├──config.py - │ ├──dataset.py - │ ├──luanch.py - │ ├──lr_generator.py - │ ├──mobilenetV2.py - ├── train.py - ├── eval.py + │ ├──config.py # parameter configuration + │ ├──dataset.py # creating dataset + │ ├──launch.py # start python script + │ ├──lr_generator.py # learning rate config + │ ├──mobilenetV3.py # MobileNetV3 architecture + ├── train.py # training script + ├── eval.py # evaluation script ``` -## Training process +## [Training process](#contents) ### Usage +You can start training using python or shell scripts. The usage of shell scripts as follows: + - GPU: sh run_trian.sh GPU [DEVICE_NUM] [VISIABLE_DEVICES(0,1,2,3,4,5,6,7)] [DATASET_PATH] ### Launch ``` # training example - GPU: sh run_train.sh GPU 8 0,1,2,3,4,5,6,7 ~/imagenet/train/ + python: + GPU: python train.py --dataset_path ~/imagenet/train/ --device_targe GPU + shell: + GPU: sh run_train.sh GPU 8 0,1,2,3,4,5,6,7 ~/imagenet/train/ ``` ### Result @@ -80,16 +100,22 @@ epoch: [ 1/200], step:[ 624/ 625], loss:[3.917/3.917], time:[138221.250], lr: epoch time: 138331.250, per step time: 221.330, avg loss: 3.917 ``` -## Eval process +## [Eval process](#contents) ### Usage +You can start training using python or shell scripts. The usage of shell scripts as follows: + - GPU: sh run_infer.sh GPU [DATASET_PATH] [CHECKPOINT_PATH] ### Launch ``` # infer example + python: + GPU: python eval.py --dataset_path ~/imagenet/val/ --checkpoint_path mobilenet_199.ckpt --device_targe GPU + + shell: GPU: sh run_infer.sh GPU ~/imagenet/val/ ~/train/mobilenet-200_625.ckpt ``` @@ -103,46 +129,50 @@ Inference result will be stored in the example path, you can find result like th result: {'acc': 0.71976314102564111} ckpt=/path/to/checkpoint/mobilenet-200_625.ckpt ``` -# Model description +# [Model description](#contents) -## Performance +## [Performance](#contents) ### Training Performance -| Parameters | MobilenetV3 | | -| -------------------------- | ---------------------------------------------------------- | ------------------------- | -| Model Version | | large | -| Resource | Ascend 910, cpu:2.60GHz 56cores, memory:314G | NV SMX2 V100-32G | -| uploaded Date | 05/06/2020 | 05/06/2020 | -| MindSpore Version | 0.3.0 | 0.3.0 | -| Dataset | ImageNet | ImageNet | -| Training Parameters | src/config.py | src/config.py | -| Optimizer | Momentum | Momentum | -| Loss Function | SoftmaxCrossEntropy | SoftmaxCrossEntropy | -| outputs | | | -| Loss | | 1.913 | -| Accuracy | | ACC1[77.57%] ACC5[92.51%] | -| Total time | | | -| Params (M) | | | -| Checkpoint for Fine tuning | | | -| Model for inference | | | +| Parameters | MobilenetV3 | +| -------------------------- | ------------------------- | +| Model Version | large | +| Resource | NV SMX2 V100-32G | +| uploaded Date | 05/06/2020 | +| MindSpore Version | 0.3.0 | +| Dataset | ImageNet | +| Training Parameters | src/config.py | +| Optimizer | Momentum | +| Loss Function | SoftmaxCrossEntropy | +| outputs | | +| Loss | 1.913 | +| Accuracy | ACC1[77.57%] ACC5[92.51%] | +| Total time | | +| Params (M) | | +| Checkpoint for Fine tuning | | +| Model for inference | | #### Inference Performance -| Parameters | | | | -| -------------------------- | ----------------------------- | ------------------------- | -------------------- | -| Model Version | V1 | | | -| Resource | Huawei 910 | NV SMX2 V100-32G | Huawei 310 | -| uploaded Date | 05/06/2020 | 05/22/2020 | | -| MindSpore Version | 0.2.0 | 0.2.0 | 0.2.0 | -| Dataset | ImageNet, 1.2W | ImageNet, 1.2W | ImageNet, 1.2W | -| batch_size | | 130(8P) | | -| outputs | | | | -| Accuracy | | ACC1[75.43%] ACC5[92.51%] | | -| Speed | | | | -| Total time | | | | -| Model for inference | | | | - - -# ModelZoo Homepage - [Link](https://gitee.com/mindspore/mindspore/tree/master/mindspore/model_zoo) \ No newline at end of file +| Parameters | | +| -------------------------- | -------------------- | +| Model Version | | +| Resource | NV SMX2 V100-32G | +| uploaded Date | 05/22/2020 | +| MindSpore Version | 0.2.0 | +| Dataset | ImageNet, 1.2W | +| batch_size | 130(8P) | +| outputs | | +| Accuracy | ACC1[75.43%] ACC5[92.51%] | +| Speed | | +| Total time | | +| Model for inference | | + +# [Description of Random Situation](#contents) + +In dataset.py, we set the seed inside “create_dataset" function. We also use random seed in train.py. + +# [ModelZoo Homepage](#contents) + +Please check the official [homepage](https://gitee.com/mindspore/mindspore/tree/master/model_zoo). diff --git a/model_zoo/official/cv/mobilenetv3/eval.py b/model_zoo/official/cv/mobilenetv3/eval.py index a44756dfef..43ebb1f21a 100644 --- a/model_zoo/official/cv/mobilenetv3/eval.py +++ b/model_zoo/official/cv/mobilenetv3/eval.py @@ -15,33 +15,26 @@ """ eval. """ -import os import argparse from mindspore import context from mindspore import nn from mindspore.train.model import Model from mindspore.train.serialization import load_checkpoint, load_param_into_net -from mindspore.common import dtype as mstype from src.dataset import create_dataset -from src.config import config_ascend, config_gpu +from src.config import config_gpu from src.mobilenetV3 import mobilenet_v3_large parser = argparse.ArgumentParser(description='Image classification') parser.add_argument('--checkpoint_path', type=str, default=None, help='Checkpoint file path') parser.add_argument('--dataset_path', type=str, default=None, help='Dataset path') -parser.add_argument('--device_target', type=str, default=None, help='run device_target') +parser.add_argument('--device_target', type=str, default="GPU", help='run device_target') args_opt = parser.parse_args() if __name__ == '__main__': config = None - if args_opt.device_target == "Ascend": - config = config_ascend - device_id = int(os.getenv('DEVICE_ID')) - context.set_context(mode=context.GRAPH_MODE, device_target="Ascend", - device_id=device_id, save_graphs=False) - elif args_opt.device_target == "GPU": + if args_opt.device_target == "GPU": config = config_gpu context.set_context(mode=context.GRAPH_MODE, device_target="GPU", save_graphs=False) @@ -52,12 +45,6 @@ if __name__ == '__main__': is_grad=False, sparse=True, reduction='mean') net = mobilenet_v3_large(num_classes=config.num_classes) - if args_opt.device_target == "Ascend": - net.to_float(mstype.float16) - for _, cell in net.cells_and_names(): - if isinstance(cell, nn.Dense): - cell.to_float(mstype.float32) - dataset = create_dataset(dataset_path=args_opt.dataset_path, do_train=False, config=config, diff --git a/model_zoo/official/cv/mobilenetv3/train.py b/model_zoo/official/cv/mobilenetv3/train.py index b103bfa360..60f3723244 100644 --- a/model_zoo/official/cv/mobilenetv3/train.py +++ b/model_zoo/official/cv/mobilenetv3/train.py @@ -13,7 +13,7 @@ # limitations under the License. # ============================================================================ """train_imagenet.""" -import os + import time import argparse import random @@ -47,20 +47,10 @@ de.config.set_seed(1) parser = argparse.ArgumentParser(description='Image classification') parser.add_argument('--dataset_path', type=str, default=None, help='Dataset path') parser.add_argument('--pre_trained', type=str, default=None, help='Pretrained checkpoint path') -parser.add_argument('--device_target', type=str, default=None, help='run device_target') +parser.add_argument('--device_target', type=str, default="GPU", help='run device_target') args_opt = parser.parse_args() -if args_opt.device_target == "Ascend": - device_id = int(os.getenv('DEVICE_ID')) - rank_id = int(os.getenv('RANK_ID')) - rank_size = int(os.getenv('RANK_SIZE')) - run_distribute = rank_size > 1 - device_id = int(os.getenv('DEVICE_ID')) - context.set_context(mode=context.GRAPH_MODE, - device_target="Ascend", - device_id=device_id, - save_graphs=False) -elif args_opt.device_target == "GPU": +if args_opt.device_target == "GPU": context.set_context(mode=context.GRAPH_MODE, device_target="GPU", save_graphs=False)