Skip to content
Snippets Groups Projects
Commit 99b17e6a authored by i-robot's avatar i-robot Committed by Gitee
Browse files

!1594 modify README in some networks 

Merge pull request !1594 from zhaoting/readme
parents be260f53 a896cda5
No related branches found
No related tags found
No related merge requests found
Hide whitespace changes
Inline Side-by-side
Showing
with 163 additions and 70 deletions
official/cv/mobilenetv2/default_config.yaml
+ 1
− 1
View file @ 99b17e6a
......@@ -43,7 +43,7 @@ activation: "Softmax"
# Image classification trian. train_parse_args():return train_args
dataset_path: "/cache/data"
pretrain_ckpt: "./mobilenetv2-200_625.ckpt"
pretrain_ckpt: ""
freeze_layer: ""
filter_head: False
enable_cache: False
......
official/cv/mobilenetv2/default_config_boost.yaml
+ 1
− 1
View file @ 99b17e6a
......@@ -43,7 +43,7 @@ activation: "Softmax"
# Image classification trian. train_parse_args():return train_args
dataset_path: "/cache/data"
pretrain_ckpt: "./mobilenetv2-200_625.ckpt"
pretrain_ckpt: ""
freeze_layer: ""
filter_head: False
enable_cache: False
......
official/cv/mobilenetv2/default_config_cpu.yaml
+ 1
− 1
View file @ 99b17e6a
......@@ -43,7 +43,7 @@ activation: "Softmax"
# Image classification trian. train_parse_args():return train_args
dataset_path: "/cache/data"
pretrain_ckpt: "./mobilenetv2-200_625.ckpt"
pretrain_ckpt: ""
freeze_layer: ""
filter_head: False
enable_cache: False
......
official/cv/mobilenetv2/default_config_gpu.yaml
+ 1
− 1
View file @ 99b17e6a
......@@ -42,7 +42,7 @@ activation: "Softmax"
# Image classification trian. train_parse_args():return train_args
dataset_path: "/cache/data"
pretrain_ckpt: "./mobilenetv2-200_625.ckpt"
pretrain_ckpt: ""
freeze_layer: ""
filter_head: False
enable_cache: False
......
official/cv/ssd/README.md
+ 72
− 33
View file @ 99b17e6a
......@@ -26,8 +26,6 @@
- [Post Training Quantization](#post-training-quantization)
- [Model Description](#model-description)
- [Performance](#performance)
- [Evaluation Performance](#evaluation-performance)
- [Inference Performance](#inference-performance)
- [Description of Random Situation](#description-of-random-situation)
- [ModelZoo Homepage](#modelzoo-homepage)
......@@ -517,7 +515,7 @@ Current batch size can only be set to 1. The precision calculation process needs
```shell
# Ascend310 inference
bash run_infer_310.sh [MINDIR_PATH] [DATA_PATH] [DVPP] [DEVICE_ID]
bash run_infer_310.sh [MINDIR_PATH] [DATA_PATH] [DVPP] [CONFIG_PATH] [DEVICE_ID]
```
- `DVPP` is mandatory, and must choose from ["DVPP", "CPU"], it's case-insensitive. Note that the image shape of ssd_vgg16 inference is [300, 300], The DVPP hardware restricts width 16-alignment and height even-alignment. Therefore, the network needs to use the CPU operator to process images.
......@@ -592,36 +590,77 @@ mAP: 0.23657619676441116
### [Performance](#contents)
#### Evaluation Performance
| Parameters | Ascend | GPU | Ascend | GPU |
| ------------------- | ----------------------------------------------------------------------------- | ----------------------------------------------------------------------------- | ----------------------------------------------------------------------------- |----------------------------------------------------------------------------- |
| Model Version | SSD V1 | SSD V1 | SSD-Mobilenet-V1-Fpn |SSD-Mobilenet-V1-Fpn |
| Resource | Ascend 910; CPU 2.60GHz, 192cores; Memory 755G; OS Euler2.8 | NV SMX2 V100-16G | Ascend 910; CPU 2.60GHz, 192cores; Memory 755G; OS Euler2.8 |NV SMX2 V100-32G |
| uploaded Date | 07/05/2021 (month/day/year) | 09/24/2020 (month/day/year) | 01/13/2021 (month/day/year) |07/20/2021 (month/day/year) |
| MindSpore Version | 1.3.0 | 1.0.0 | 1.1.0 |1.3.0 |
| Dataset | COCO2017 | COCO2017 | COCO2017 |COCO2017 |
| Training Parameters | epoch = 500, batch_size = 32 | epoch = 800, batch_size = 32 | epoch = 60, batch_size = 32 | epoch = 60, batch_size = 16 |
| Optimizer | Momentum | Momentum | Momentum |Momentum |
| Loss Function | Sigmoid Cross Entropy,SmoothL1Loss | Sigmoid Cross Entropy,SmoothL1Loss | Sigmoid Cross Entropy,SmoothL1Loss |Sigmoid Cross Entropy,SmoothL1Loss |
| Speed | 8pcs: 90ms/step | 8pcs: 121ms/step | 8pcs: 547ms/step |1pcs: 547ms/step |
| Total time | 8pcs: 4.81hours | 8pcs: 12.31hours | 8pcs: 4.22hours |1pcs: 4.22hours |
| Parameters (M) | 34 | 34 | 48M |97M |
| Scripts | <https://gitee.com/mindspore/models/tree/master/official/cv/ssd> | <https://gitee.com/mindspore/models/tree/master/official/cv/ssd> | <https://gitee.com/mindspore/models/tree/master/official/cv/ssd> |<https://gitee.com/mindspore/models/tree/master/official/cv/ssd> |
#### Inference Performance
| Parameters | Ascend | GPU | Ascend | GPU |
| ------------------- | --------------------------- | --------------------------- | --------------------------- |--------------------------- |
| Model Version | SSD V1 | SSD V1 | SSD-Mobilenet-V1-Fpn | SSD-Mobilenet-V1-Fpn |
| Resource | Ascend 910; OS Euler2.8 | GPU |Ascend 910; OS Euler2.8 | NV SMX2 V100-32G |
| Uploaded Date | 07/05/2020 (month/day/year) | 09/24/2020 (month/day/year) | 09/24/2020 (month/day/year) | 07/20/2021 (month/day/year) |
| MindSpore Version | 1.3.0 | 1.0.0 | 1.1.0 | 1.3.0 |
| Dataset | COCO2017 | COCO2017 | COCO2017 | COCO2017 |
| batch_size | 1 | 1 | 1 | 1 |
| outputs | mAP | mAP | mAP | mAP |
| Accuracy | IoU=0.50: 23.8% | IoU=0.50: 22.4% | Iout=0.50: 30% | Iout=0.50: 30% |
| Model for inference | 34M(.ckpt file) | 34M(.ckpt file) | 48M(.ckpt file) | 97M(.ckpt file) |
| Parameters | Ascend | GPU |
| ------------------- | --------------------------- | --------------------------- |
| Model Version | SSD MobileNetV2 | SSD MobileNetV2 |
| Resource | Ascend 910; OS Euler2.8 | GPU(Tesla V100 PCIE)，CPU 2.1GHz 64 cores，Memory 128G |
| Uploaded Date | 07/05/2020 (month/day/year) | 09/24/2020 (month/day/year) |
| MindSpore Version | 1.3.0 | 1.3.0 |
| Dataset | COCO2017 | COCO2017 |
| Training Parameters | epoch = 500, batch_size = 32 | epoch = 800, batch_size = 24(8ps)/32(1ps) |
| Optimizer | Momentum | Momentum |
| Loss Function | Sigmoid Cross Entropy,SmoothL1Loss | Sigmoid Cross Entropy,SmoothL1Loss |
| Speed | 8pcs: 90ms/step | 8pcs: 121ms/step |
| Total time | 8pcs: 4.81hours | 8pcs: 12.31hours |
| outputs | mAP | mAP |
| Accuracy | IoU=0.50: 22% | IoU=0.50: 22% |
| Model for inference | 34M(.ckpt file) | 34M(.ckpt file) |
| configuration | ssd300_config.yaml |ssd300_config_gpu.yaml |
| Scripts | <https://gitee.com/mindspore/models/tree/master/official/cv/ssd> |
| Parameters | Ascend | GPU |
| ------------------- | --------------------------- | --------------------------- |
| Model Version | SSD-MobileNetV1-FPN | SSD-MobileNetV1-FPN |
| Resource | Ascend 910; OS Euler2.8 | GPU(Tesla V100 PCIE)，CPU 2.1GHz 64 cores，Memory 128G |
| Uploaded Date | 11/14/2020 (month/day/year) | 07/23/2021 (month/day/year) |
| MindSpore Version | 1.3.0 | 1.3.0 |
| Dataset | COCO2017 | COCO2017 |
| Training Parameters | epoch = 60, batch_size = 32 | epoch = 60, batch_size = 16 |
| Optimizer | Momentum | Momentum |
| Loss Function | Sigmoid Cross Entropy,SmoothL1Loss | Sigmoid Cross Entropy,SmoothL1Loss |
| Speed | 8pcs: 408 ms/step | 8pcs: 640 ms/step |
| Total time | 8pcs: 4.5 hours | 8pcs: 9.7 hours |
| outputs | mAP | mAP |
| Accuracy | IoU=0.50: 29.1 % | IoU=0.50: 29.1 % |
| Model for inference | 96M(.ckpt file) | 96M(.ckpt file) |
| configuration | ssd_mobilenet_v1_fpn_config.yaml |ssd_mobilenet_v1_fpn_config_gpu.yaml |
| Scripts | <https://gitee.com/mindspore/models/tree/master/official/cv/ssd> |
| Parameters | Ascend | GPU |
| ------------------- | --------------------------- | --------------------------- |
| Model Version | SSD-Resnet50-FPN | SSD-Resnet50-FPN |
| Resource | Ascend 910; OS Euler2.8 | GPU(Tesla V100 PCIE)，CPU 2.1GHz 64 cores，Memory 128G |
| Uploaded Date | 03/10/2021 (month/day/year) | 07/23/2021 (month/day/year) |
| MindSpore Version | 1.3.0 | 1.3.0 |
| Dataset | COCO2017 | COCO2017 |
| Training Parameters | epoch = 60, batch_size = 32 | epoch = 60, batch_size = 16 |
| Optimizer | Momentum | Momentum |
| Loss Function | Sigmoid Cross Entropy,SmoothL1Loss | Sigmoid Cross Entropy,SmoothL1Loss |
| Speed | 8pcs: 345 ms/step | 8pcs: 877 ms/step |
| Total time | 8pcs: 4.1 hours | 8pcs: 12 hours |
| outputs | mAP | mAP |
| Accuracy | IoU=0.50: 34.3% | IoU=0.50: 34.3 % |
| Model for inference | 255M(.ckpt file) | 255M(.ckpt file) |
| configuration | ssd_resnet50_fpn_config.yaml | ssd_resnet50_fpn_config_gpu.yaml |
| Scripts | <https://gitee.com/mindspore/models/tree/master/official/cv/ssd> |
| Parameters | Ascend | GPU |
| ------------------- | --------------------------- | --------------------------- |
| Model Version | SSD VGG16 | SSD VGG16 |
| Resource | Ascend 910; OS Euler2.8 | GPU(Tesla V100 PCIE)，CPU 2.1GHz 64 cores，Memory 128G |
| Uploaded Date | 03/27/2021 (month/day/year) | 07/23/2021 (month/day/year) |
| MindSpore Version | 1.3.0 | 1.3.0 |
| Dataset | COCO2017 | COCO2017 |
| Training Parameters | epoch = 150, batch_size = 32 | epoch = 150, batch_size = 32 |
| Optimizer | Momentum | Momentum |
| Loss Function | Sigmoid Cross Entropy,SmoothL1Loss | Sigmoid Cross Entropy,SmoothL1Loss |
| Speed | 8pcs: 117 ms/step | 8pcs: 403 ms/step |
| Total time | 8pcs: 4.81hours | 8pcs: 16.8 hours |
| outputs | mAP | mAP |
| Accuracy | IoU=0.50: 23.2% | IoU=0.50: 23.2% |
| Model for inference | 186M(.ckpt file) | 186M(.ckpt file) |
| configuration | ssd_vgg16_config.yaml | ssd_vgg16_config_gpu.yaml |
| Scripts | <https://gitee.com/mindspore/models/tree/master/official/cv/ssd> |
## [Description of Random Situation](#contents)
......
official/cv/ssd/README_CN.md
+ 71
− 31
View file @ 99b17e6a
......@@ -24,8 +24,6 @@
- [训练后量化推理](#训练后量化推理)
- [模型描述](#模型描述)
- [性能](#性能)
- [评估性能](#评估性能)
- [推理性能](#推理性能)
- [随机情况说明](#随机情况说明)
- [ModelZoo主页](#modelzoo主页)
......@@ -439,7 +437,8 @@ ModelArts导出mindir
```shell
# Ascend310 inference
bash run_infer_310.sh [MINDIR_PATH] [DATA_PATH] [DVPP] [DEVICE_ID] [CONFIG_PATH]
bash run_infer_310.sh [MINDIR_PATH] [DATA_PATH] [DVPP] [CONFIG_PATH] [DEVICE_ID]
```
- `DVPP` 为必填项，需要在["DVPP", "CPU"]选择，大小写均可。需要注意的是ssd_vgg16执行推理的图片尺寸为[300, 300]，由于DVPP硬件限制宽为16整除，高为2整除，因此，这个网络需要通过CPU算子对图像进行前处理。
......@@ -513,36 +512,77 @@ mAP: 0.23657619676441116
## 性能
### 评估性能
| 参数 | Ascend | GPU |
| -------------------------- | -------------------------------------------------------------| -------------------------------------------------------------|
| 模型版本 | SSD V1 | SSD V1 |
| 资源 | Ascend 910；CPU 2.60GHz，192核；内存 755GB；系统 Euler2.8 | NV SMX2 V100-16G |
| 上传日期 | 2021-07-05 | 2020-09-24 |
| MindSpore版本 | 1.3.0 | 1.0.0 |
| 数据集 | COCO2017 | COCO2017 |
| 训练参数 | epoch = 500, batch_size = 32 | epoch = 800, batch_size = 32 |
| 优化器 | Momentum | Momentum |
| 损失函数 | Sigmoid交叉熵，SmoothL1Loss | Sigmoid交叉熵，SmoothL1Loss |
| 速度 | 8卡：90毫秒/步 | 8卡：121毫秒/步 |
| 总时长 | 8卡：4.81小时 | 8卡：12.31小时 |
| 参数(M) | 34 | 34 |
|脚本 | https://gitee.com/mindspore/models/tree/master/official/cv/ssd | https://gitee.com/mindspore/models/tree/master/official/cv/ssd |
### 推理性能
| 参数 | Ascend | GPU |
| ------------------- | --------------------------- | --------------------------- |
| 网络 | SSD MobileNetV2 | SSD MobileNetV2 |
| 资源 | Ascend 910; OS Euler2.8 | GPU(Tesla V100 PCIE)，CPU 2.1GHz 64 cores，Memory 128G |
| 上传日期 | 07/05/2020 (month/day/year) | 09/24/2020 (month/day/year) |
| 昇思版本 | 1.3.0 | 1.3.0 |
| 数据集 | COCO2017 | COCO2017 |
| 训练参数 | epoch = 500, batch_size = 32 | epoch = 800, batch_size = 24(8ps)/32(1ps) |
| 优化器 | Momentum | Momentum |
| 损失函数 | Sigmoid Cross Entropy,SmoothL1Loss | Sigmoid Cross Entropy,SmoothL1Loss |
| 性能 | 8pcs: 90ms/step | 8pcs: 121ms/step |
| 训练耗时 | 8pcs: 4.81hours | 8pcs: 12.31hours |
| 推理输出 | mAP | mAP |
| 评价指标 | IoU=0.50: 22% | IoU=0.50: 22% |
| 推理模型大小 | 34M(.ckpt file) | 34M(.ckpt file) |
| 参数文件 | ssd300_config.yaml |ssd300_config_gpu.yaml |
| 脚本链接 | <https://gitee.com/mindspore/models/tree/master/official/cv/ssd> |
| 参数 | Ascend | GPU |
| ------------------- | --------------------------- | --------------------------- |
| 网络 | SSD-MobileNetV1-FPN | SSD-MobileNetV1-FPN |
| 资源 | Ascend 910; OS Euler2.8 | GPU(Tesla V100 PCIE)，CPU 2.1GHz 64 cores，Memory 128G |
| 上传日期 | 11/14/2020 (month/day/year) | 07/23/2021 (month/day/year) |
| 昇思版本 | 1.3.0 | 1.3.0 |
| 数据集 | COCO2017 | COCO2017 |
| 训练参数 | epoch = 60, batch_size = 32 | epoch = 60, batch_size = 16 |
| 优化器 | Momentum | Momentum |
| 损失函数 | Sigmoid Cross Entropy,SmoothL1Loss | Sigmoid Cross Entropy,SmoothL1Loss |
| 性能 | 8pcs: 408 ms/step | 8pcs: 640 ms/step |
| 训练耗时 | 8pcs: 4.5 hours | 8pcs: 9.7 hours |
| 推理输出 | mAP | mAP |
| 评价指标 | IoU=0.50: 29.1 % | IoU=0.50: 29.1 % |
| 推理模型大小 | 96M(.ckpt file) | 96M(.ckpt file) |
| 参数文件 | ssd_mobilenet_v1_fpn_config.yaml |ssd_mobilenet_v1_fpn_config_gpu.yaml |
| 脚本链接 | <https://gitee.com/mindspore/models/tree/master/official/cv/ssd> |
| 参数 | Ascend | GPU |
| ------------------- | --------------------------- | --------------------------- |
| 网络 | SSD-Resnet50-FPN | SSD-Resnet50-FPN |
| 资源 | Ascend 910; OS Euler2.8 | GPU(Tesla V100 PCIE)，CPU 2.1GHz 64 cores，Memory 128G |
| 上传日期 | 03/10/2021 (month/day/year) | 07/23/2021 (month/day/year) |
| 昇思版本 | 1.3.0 | 1.3.0 |
| 数据集 | COCO2017 | COCO2017 |
| 训练参数 | epoch = 60, batch_size = 32 | epoch = 60, batch_size = 16 |
| 优化器 | Momentum | Momentum |
| 损失函数 | Sigmoid Cross Entropy,SmoothL1Loss | Sigmoid Cross Entropy,SmoothL1Loss |
| 性能 | 8pcs: 345 ms/step | 8pcs: 877 ms/step |
| 训练耗时 | 8pcs: 4.1 hours | 8pcs: 12 hours |
| 推理输出 | mAP | mAP |
| 评价指标 | IoU=0.50: 34.3% | IoU=0.50: 34.3 % |
| 推理模型大小 | 255M(.ckpt file) | 255M(.ckpt file) |
| 参数文件 | ssd_resnet50_fpn_config.yaml | ssd_resnet50_fpn_config_gpu.yaml |
| 脚本链接 | <https://gitee.com/mindspore/models/tree/master/official/cv/ssd> |
| 参数 | Ascend | GPU |
| ------------------- | ----------------------------| ----------------------------|
| 模型版本 | SSD V1 | SSD V1 |
| 资源 | Ascend 910；系统 Euler2.8 | GPU |
| 上传日期 | 2021-07-05 | 2020-09-24 |
| MindSpore版本 | 1.3.0 | 1.0.0 |
| 数据集 | COCO2017 | COCO2017 |
| batch_size | 1 | 1 |
| 输出 | mAP | mAP |
| 准确率 | IoU=0.50: 23.8% | IoU=0.50: 22.4% |
| 推理模型 | 34M（.ckpt文件） | 34M（.ckpt文件） |
| ------------------- | --------------------------- | --------------------------- |
| 网络 | SSD VGG16 | SSD VGG16 |
| 资源 | Ascend 910; OS Euler2.8 | GPU(Tesla V100 PCIE)，CPU 2.1GHz 64 cores，Memory 128G |
| 上传日期 | 03/27/2021 (month/day/year) | 07/23/2021 (month/day/year) |
| 昇思版本 | 1.3.0 | 1.3.0 |
| 数据集 | COCO2017 | COCO2017 |
| 训练参数 | epoch = 150, batch_size = 32 | epoch = 150, batch_size = 32 |
| 优化器 | Momentum | Momentum |
| 损失函数 | Sigmoid Cross Entropy,SmoothL1Loss | Sigmoid Cross Entropy,SmoothL1Loss |
| 性能 | 8pcs: 117 ms/step | 8pcs: 403 ms/step |
| 训练耗时 | 8pcs: 4.81hours | 8pcs: 16.8 hours |
| 推理输出 | mAP | mAP |
| 评价指标 | IoU=0.50: 23.2% | IoU=0.50: 23.2% |
| 推理模型大小 | 186M(.ckpt file) | 186M(.ckpt file) |
| 参数文件 | ssd_vgg16_config.yaml | ssd_vgg16_config_gpu.yaml |
| 脚本链接 | <https://gitee.com/mindspore/models/tree/master/official/cv/ssd> |
# 随机情况说明
......
official/cv/unet/README.md
+ 4
− 0
View file @ 99b17e6a
......@@ -471,8 +471,10 @@ The above python command will run in the background. You can view the results th
| Loss | 0.22070312 | 0.21425568 |
| Speed | 1pc: 267 ms/step; | 1pc: 423 ms/step; |
| Total time | 1pc: 2.67 mins; | 1pc: 5.64 mins; |
| Accuracy | IOU 90% | IOU 90% |
| Parameters (M) | 93M | 93M |
| Checkpoint for Fine tuning | 355.11M (.ckpt file) | 355.11M (.ckpt file) |
| configuration | unet_medical_config.yaml | unet_medical_gpu_config.yaml |
| Scripts | [unet script](https://gitee.com/mindspore/models/tree/master/official/cv/unet) | [unet script](https://gitee.com/mindspore/models/tree/master/official/cv/unet) |
| Parameters | Ascend | GPU |
......@@ -489,9 +491,11 @@ The above python command will run in the background. You can view the results th
| probability | cross valid dice coeff is 0.966, cross valid IOU is 0.936 | cross valid dice coeff is 0.976,cross valid IOU is 0.955 |
| Loss | <0.1 | <0.1 |
| Speed | 1pc: 150~200 fps | 1pc：230~280 fps, 8pc：(170~210)*8 fps |
| Accuracy | IOU 93% | IOU 92% |
| Total time | 1pc: 10.8min | 1pc：8min |
| Parameters (M) | 27M | 27M |
| Checkpoint for Fine tuning | 103.4M(.ckpt file) | 103.4M(.ckpt file) |
| configuration | unet_nested_cell_config.yaml | unet_nested_cell_config.yaml|
| Scripts | [unet script](https://gitee.com/mindspore/models/tree/master/official/cv/unet) | [unet script](https://gitee.com/mindspore/models/tree/master/official/cv/unet) |
## [How to use](#contents)
......
official/cv/unet/README_CN.md
+ 5
− 1
View file @ 99b17e6a
......@@ -470,9 +470,11 @@ bash scripts/run_distribute_train_gpu.sh [RANKSIZE] [DATASET] [CONFIG_PATH]
| 损失 | 0.22070312 | 0.21425568 |
| 速度 | 1卡：267毫秒/步；8卡：280毫秒/步 | 1卡：423毫秒/步；8卡：128毫秒/步 |
| 总时长 | 1卡：2.67分钟；8卡：1.40分钟 | 1卡：5.64分钟；8卡：3.41分钟 |
| 精度 | IOU 90% | IOU 90% |
| 参数(M) | 93M | 93M |
| 微调检查点 | 355.11M (.ckpt文件) | 355.11M (.ckpt文件) |
| 脚本 | [U-Net脚本](https://gitee.com/mindspore/models/tree/master/official/cv/unet) | [U-Net脚本](https://gitee.com/mindspore/models/tree/master/official/cv/unet) |
| 配置文件 | unet_medical_config.yaml | unet_medical_gpu_config.yaml |
| 脚本| [U-Net脚本](https://gitee.com/mindspore/models/tree/master/official/cv/unet) | [U-Net脚本](https://gitee.com/mindspore/models/tree/master/official/cv/unet) |
| 参数 | Ascend | GPU |
| ----- | ------ | ----- |
......@@ -489,8 +491,10 @@ bash scripts/run_distribute_train_gpu.sh [RANKSIZE] [DATASET] [CONFIG_PATH]
| 损失 | <0.1 | <0.1 |
| 速度 | 1卡：150~200 fps | 1卡：230~280 fps, 8卡：(170~210)*8 fps|
| 总时长 | 1卡: 10.8分钟 | 1卡: 8分钟 |
| 精度 | IOU 93% | IOU 92% |
| 参数(M) | 27M | 27M |
| 微调检查点 | 103.4M(.ckpt文件) | 103.4M(.ckpt文件) |
| 配置文件 | unet_nested_cell_config.yaml | unet_nested_cell_config.yaml|
| 脚本 | [U-Net脚本](https://gitee.com/mindspore/models/tree/master/official/cv/unet) | [U-Net脚本](https://gitee.com/mindspore/models/tree/master/official/cv/unet) |
### 用法
......
research/cv/ssd_ghostnet/README.md
+ 7
− 1
View file @ 99b17e6a
......@@ -106,8 +106,14 @@ python train.py
# distributed training on Ascend
bash run_distribute_train_ghostnet.sh [DEVICE_NUM] [EPOCH_SIZE] [LR] [DATASET] [RANK_TABLE_FILE]
# distributed training on GPU
bash run_distribute_train_gpu.sh [DEVICE_NUM] [EPOCH_SIZE] [LR] [DATASET] [CONFIG_PATH] [PRE_TRAINED](optional) [PRE_TRAINED_EPOCH_SIZE](optional)
# run eval on Ascend
python eval.py --device_id 0 --dataset coco --checkpoint_file_path LOG4/ssd-500_458.ckpt
# run eval on GPU
python eval.py --device_id 0 --dataset coco --checkpoint_path LOG4/ssd-500_458.ckpt --device_target="GPU"
```
If you want to run in modelarts, please check the official documentation of [modelarts](https://support.huaweicloud.com/modelarts/), and you can start training and evaluation as follows:
......@@ -366,5 +372,5 @@ mAP: 0.24270569394180577
| Dataset | COCO2017 |
| batch_size | 1 |
| outputs | mAP |
| Accuracy | IoU=0.50: 24.2% |
| Accuracy | IoU=0.50: 24.1% |
| Model for inference | 52.5M(.ckpt file) |
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment