"backend/vscode:/vscode.git/clone" did not exist on "7e5e2c42c942a52388ca73bb1edc3ffaaadf76a6"
Commit 93c5eac2 authored by chenych's avatar chenych
Browse files

update val

parent 1bb2aaec
......@@ -144,12 +144,6 @@ python test.py --pre_trained_model <checkpoint path> --coco_path <coco path>
其余对应参数与训练模型参数需一致, 详情请参考代码里面的参数配置:
#### 单卡推理
```
bash val.sh
```
#### 多卡推理
```
......@@ -166,9 +160,10 @@ COCO2017测试集上的单张图像结果展示:
### 精度
在COCO2017的测试集上进行单卡测试, 结果如下表所示:
在COCO2017的测试集上进行单卡测试, 测试模型config: ./configs/two_stage/deformable-detr-hybrid-branch/12eps/r50_hybrid_branch_lambda1_group6_t1500_dp0_mqs_lft_deformable_detr_plus_iterative_bbox_refinement_plus_plus_two_stage.sh
结果如下表所示:
根据测试结果情况填写表格:
| Name | Backbone | query | epochs | AP |
| :--------: | :------: | :------: | :------: | :------: |
| H-Deformable-DETR + tricks(our) | R50 | 300 | 12 | xxx |
......
#!/usr/bin/env bash
# coco_path是训练数据集地址,数据是coco format
sh <config path> \
--coco_path <coco path> \
--resume <checkpoint path> \
--eval
#!/usr/bin/env bash
export HIP_VISIBLE_DEVICES=0,1,2,3 # 自行修改为训练的卡号和数量
export HIP_VISIBLE_DEVICES=0,1,2,3 # 自行修改为训练的卡号和数量, 单卡可设置一张卡,下面的参数对应调整
export HSA_FORCE_FINE_GRAIN_PCIE=1
export USE_MIOPEN_BATCHNORM=1
......
Markdown is supported
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