提交 6d04e1a8 编写于 作者: M mindspore-ci-bot 提交者: Gitee

!2115 change readme.md

Merge pull request !2115 from chenzhongming/r0.3
...@@ -4,7 +4,7 @@ MobileNetV2 is a significant improvement over MobileNetV1 and pushes the state o ...@@ -4,7 +4,7 @@ MobileNetV2 is a significant improvement over MobileNetV1 and pushes the state o
MobileNetV2 builds upon the ideas from MobileNetV1, using depthwise separable convolution as efficient building blocks. However, V2 introduces two new features to the architecture: 1) linear bottlenecks between the layers, and 2) shortcut connections between the bottlenecks1. MobileNetV2 builds upon the ideas from MobileNetV1, using depthwise separable convolution as efficient building blocks. However, V2 introduces two new features to the architecture: 1) linear bottlenecks between the layers, and 2) shortcut connections between the bottlenecks1.
[Paper](https://arxiv.org/pdf/1801.04381) 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. [Paper](https://arxiv.org/pdf/1801.04381) Sandler, Mark, et al. "Mobilenetv2: Inverted residuals and linear bottlenecks." Proceedings of the IEEE conference on computer vision and pattern recognition. 2018.
# Dataset # Dataset
......
...@@ -4,7 +4,7 @@ MobileNetV2 is a significant improvement over MobileNetV1 and pushes the state o ...@@ -4,7 +4,7 @@ MobileNetV2 is a significant improvement over MobileNetV1 and pushes the state o
MobileNetV2 builds upon the ideas from MobileNetV1, using depthwise separable convolution as efficient building blocks. However, V2 introduces two new features to the architecture: 1) linear bottlenecks between the layers, and 2) shortcut connections between the bottlenecks1. MobileNetV2 builds upon the ideas from MobileNetV1, using depthwise separable convolution as efficient building blocks. However, V2 introduces two new features to the architecture: 1) linear bottlenecks between the layers, and 2) shortcut connections between the bottlenecks1.
[Paper](https://arxiv.org/pdf/1801.04381) 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. [Paper](https://arxiv.org/pdf/1801.04381) Sandler, Mark, et al. "Mobilenetv2: Inverted residuals and linear bottlenecks." Proceedings of the IEEE conference on computer vision and pattern recognition. 2018.
# Dataset # Dataset
...@@ -16,7 +16,6 @@ Dataset used: imagenet ...@@ -16,7 +16,6 @@ Dataset used: imagenet
- Data format: RGB images. - Data format: RGB images.
- Note: Data will be processed in src/dataset.py - Note: Data will be processed in src/dataset.py
# Environment Requirements # Environment Requirements
- Hardware(Ascend) - Hardware(Ascend)
...@@ -48,6 +47,8 @@ Dataset used: imagenet ...@@ -48,6 +47,8 @@ Dataset used: imagenet
├── eval.py ├── eval.py
``` ```
Notation: Current hyperparameters only test on 4 cards while training, if want to use 8 cards for training, should change parameters like learning rate in 'src/config.py'.
## Training process ## Training process
### Usage ### Usage
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册