Skip to content

Commit c616c10

Browse files
yghstillyghstill
authored
fix new PicoDet name (PaddlePaddle#5414)
1 parent 7018dad commit c616c10

12 files changed

+70
-56
lines changed

‎configs/picodet/README.md

+29-28
Original file line numberDiff line numberDiff line change
@@ -6,7 +6,12 @@ English | [简体中文](README_cn.md)
66

77
## News
88

9-
- Released a new series of PP-PicoDet models, with greatly improved accuracy and optimized CPU prediction speed. **(2022.03.20)**
9+
- Released a new series of PP-PicoDet models, it was used TAL/Task-aligned-Head and optimized PAN, which improved the accuracy and optimized CPU prediction speed. Moreover the training speed is greatly improved. **(2022.03.20)**
10+
11+
### Legacy Model
12+
13+
- Please refer to: [PicoDet 2021.10版本](./legacy_model/)
14+
1015
## Introduction
1116

1217
We developed a series of lightweight models, named `PP-PicoDet`. Because of the excellent performance, our models are very suitable for deployment on mobile or CPU. For more details, please refer to our [report on arXiv](https://arxiv.org/abs/2111.00902).
@@ -25,15 +30,15 @@ We developed a series of lightweight models, named `PP-PicoDet`. Because of the
2530

2631
| Model | Input size | mAP<sup>val<br>0.5:0.95 | mAP<sup>val<br>0.5 | Params<br><sup>(M) | FLOPS<br><sup>(G) | Latency<sup><small>[NCNN](#latency)</small><sup><br><sup>(ms) | Latency<sup><small>[Lite](#latency)</small><sup><br><sup>(ms) | Download | Config |
2732
| :-------- | :--------: | :---------------------: | :----------------: | :----------------: | :---------------: | :-----------------------------: | :-----------------------------: | :----------------------------------------: | :--------------------------------------- |
28-
| PicoDet-XS | 320*320 | 23.5 | 36.1 | - | - | - | - | [model](https://paddledet.bj.bcebos.com/models/picodetv2/picodet_xs_320_coco.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/picodetv2/train_picodet_xs_320_coco.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_xs_320_coco.yml) |
29-
| PicoDet-XS | 416*416 | 26.2 | 39.3 | - | - | - | - | [model](https://paddledet.bj.bcebos.com/models/picodetv2/picodet_xs_416_coco.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/picodetv2/train_picodet_xs_416_coco.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_xs_416_coco.yml) |
30-
| PicoDet-S | 320*320 | 29.1 | 43.4 | - | - | - | - | [model](https://paddledet.bj.bcebos.com/models/picodetv2/picodet_s_320_coco.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/picodetv2/train_picodet_s_320_coco.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_s_320_coco.yml) |
31-
| PicoDet-S | 416*416 | 32.5 | 47.6 | - | - | - | - | [model](https://paddledet.bj.bcebos.com/models/picodetv2/picodet_s_416_coco.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/picodetv2/train_picodet_s_416_coco.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_s_416_coco.yml) |
32-
| PicoDet-M | 320*320 | 34.4 | 50.0 | - | - | - | - | [model](https://paddledet.bj.bcebos.com/models/picodetv2/picodet_m_320_coco.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/picodetv2/train_picodet_m_320_coco.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_m_320_coco.yml) |
33-
| PicoDet-M | 416*416 | 37.5 | 53.4 | - | - | - | - | [model](https://paddledet.bj.bcebos.com/models/picodetv2/picodet_m_416_coco.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/picodetv2/train_picodet_m_416_coco.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_m_416_coco.yml) |
34-
| PicoDet-L | 320*320 | 36.1 | 52.0 | - | - | - | - | [model](https://paddledet.bj.bcebos.com/models/picodetv2/picodet_l_320_coco.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/picodetv2/train_picodet_l_320_coco.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_l_320_coco.yml) |
35-
| PicoDet-L | 416*416 | 39.4 | 55.7 | - | - | - | - | [model](https://paddledet.bj.bcebos.com/models/picodetv2/picodet_l_416_coco.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/picodetv2/train_picodet_l_416_coco.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_l_416_coco.yml) |
36-
| PicoDet-L | 640*640 | 42.3 | 59.2 | - | - | - | - | [model](https://paddledet.bj.bcebos.com/models/picodetv2/picodet_l_640_coco.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/picodetv2/train_picodet_l_640_coco.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_l_640_coco.yml) |
33+
| PicoDet-XS | 320*320 | 23.5 | 36.1 | - | - | - | - | [model](https://paddledet.bj.bcebos.com/models/picodet_xs_320_coco_lcnet.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/train_picodet_xs_320_coco_lcnet.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_xs_320_coco_lcnet.yml) |
34+
| PicoDet-XS | 416*416 | 26.2 | 39.3 | - | - | - | - | [model](https://paddledet.bj.bcebos.com/models/picodet_xs_416_coco_lcnet.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/train_picodet_xs_416_coco_lcnet.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_xs_416_coco_lcnet.yml) |
35+
| PicoDet-S | 320*320 | 29.1 | 43.4 | - | - | - | - | [model](https://paddledet.bj.bcebos.com/models/picodet_s_320_coco_lcnet.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/train_picodet_s_320_coco_lcnet.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_s_320_coco_lcnet.yml) |
36+
| PicoDet-S | 416*416 | 32.5 | 47.6 | - | - | - | - | [model](https://paddledet.bj.bcebos.com/models/picodet_s_416_coco_lcnet.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/train_picodet_s_416_coco_lcnet.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_s_416_coco_lcnet.yml) |
37+
| PicoDet-M | 320*320 | 34.4 | 50.0 | - | - | - | - | [model](https://paddledet.bj.bcebos.com/models/picodet_m_320_coco_lcnet.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/train_picodet_m_320_coco_lcnet.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_m_320_coco_lcnet.yml) |
38+
| PicoDet-M | 416*416 | 37.5 | 53.4 | - | - | - | - | [model](https://paddledet.bj.bcebos.com/models/picodet_m_416_coco_lcnet.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/train_picodet_m_416_coco_lcnet.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_m_416_coco_lcnet.yml) |
39+
| PicoDet-L | 320*320 | 36.1 | 52.0 | - | - | - | - | [model](https://paddledet.bj.bcebos.com/models/picodet_l_320_coco_lcnet.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/train_picodet_l_320_coco_lcnet.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_l_320_coco_lcnet.yml) |
40+
| PicoDet-L | 416*416 | 39.4 | 55.7 | - | - | - | - | [model](https://paddledet.bj.bcebos.com/models/picodet_l_416_coco_lcnet.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/train_picodet_l_416_coco_lcnet.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_l_416_coco_lcnet.yml) |
41+
| PicoDet-L | 640*640 | 42.3 | 59.2 | - | - | - | - | [model](https://paddledet.bj.bcebos.com/models/picodet_l_640_coco_lcnet.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/train_picodet_l_640_coco_lcnet.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_l_640_coco_lcnet.yml) |
3742

3843
<details open>
3944
<summary><b>Table Notes:</b></summary>
@@ -44,10 +49,6 @@ We developed a series of lightweight models, named `PP-PicoDet`. Because of the
4449

4550
</details>
4651

47-
### Legacy Model
48-
49-
- Please refer to: [PicoDet 2021.10版本](./legacy_model/)
50-
5152
#### Benchmark of Other Models
5253

5354
| Model | Input size | mAP<sup>val<br>0.5:0.95 | mAP<sup>val<br>0.5 | Params<br><sup>(M) | FLOPS<br><sup>(G) | Latency<sup><small>[NCNN](#latency)</small><sup><br><sup>(ms) |
@@ -90,7 +91,7 @@ We developed a series of lightweight models, named `PP-PicoDet`. Because of the
9091
```shell
9192
# training on single-GPU
9293
export CUDA_VISIBLE_DEVICES=0
93-
python tools/train.py -c configs/picodet/picodet_s_320_coco.yml --eval
94+
python tools/train.py -c configs/picodet/picodet_s_320_coco_lcnet.yml --eval
9495
```
9596

9697
- Training model on multi-GPU:
@@ -99,21 +100,21 @@ python tools/train.py -c configs/picodet/picodet_s_320_coco.yml --eval
99100
```shell
100101
# training on single-GPU
101102
export CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7
102-
python -m paddle.distributed.launch --gpus 0,1,2,3,4,5,6,7 tools/train.py -c configs/picodet/picodet_s_320_coco.yml --eval
103+
python -m paddle.distributed.launch --gpus 0,1,2,3,4,5,6,7 tools/train.py -c configs/picodet/picodet_s_320_coco_lcnet.yml --eval
103104
```
104105

105106
- Evaluation:
106107

107108
```shell
108-
python tools/eval.py -c configs/picodet/picodet_s_320_coco.yml \
109-
-o weights=https://paddledet.bj.bcebos.com/models/picodetv2/picodet_s_320_coco.pdparams
109+
python tools/eval.py -c configs/picodet/picodet_s_320_coco_lcnet.yml \
110+
-o weights=https://paddledet.bj.bcebos.com/models/picodet_s_320_coco_lcnet.pdparams
110111
```
111112

112113
- Infer:
113114

114115
```shell
115-
python tools/infer.py -c configs/picodet/picodet_s_320_coco.yml \
116-
-o weights=https://paddledet.bj.bcebos.com/models/picodetv2/picodet_s_320_coco.pdparams
116+
python tools/infer.py -c configs/picodet/picodet_s_320_coco_lcnet.yml \
117+
-o weights=https://paddledet.bj.bcebos.com/models/picodet_s_320_coco_lcnet.pdparams
117118
```
118119

119120
Detail also can refer to [Quick start guide](https://github.com/PaddlePaddle/PaddleDetection/blob/develop/docs/tutorials/GETTING_STARTED.md).
@@ -130,8 +131,8 @@ Detail also can refer to [Quick start guide](https://github.com/PaddlePaddle/Pad
130131

131132
```shell
132133
cd PaddleDetection
133-
python tools/export_model.py -c configs/picodet/picodet_s_320_coco.yml \
134-
-o weights=https://paddledet.bj.bcebos.com/models/picodetv2/picodet_s_320_coco.pdparams \
134+
python tools/export_model.py -c configs/picodet/picodet_s_320_coco_lcnet.yml \
135+
-o weights=https://paddledet.bj.bcebos.com/models/picodet_s_320_coco_lcnet.pdparams \
135136
--output_dir=inference_model
136137
```
137138

@@ -150,9 +151,9 @@ pip install paddlelite
150151

151152
```shell
152153
# FP32
153-
paddle_lite_opt --model_dir=inference_model/picodet_s_320_coco --valid_targets=arm --optimize_out=picodet_s_320_coco_fp32
154+
paddle_lite_opt --model_dir=inference_model/picodet_s_320_coco_lcnet --valid_targets=arm --optimize_out=picodet_s_320_coco_fp32
154155
# FP16
155-
paddle_lite_opt --model_dir=inference_model/picodet_s_320_coco --valid_targets=arm --optimize_out=picodet_s_320_coco_fp16 --enable_fp16=true
156+
paddle_lite_opt --model_dir=inference_model/picodet_s_320_coco_lcnet --valid_targets=arm --optimize_out=picodet_s_320_coco_fp16 --enable_fp16=true
156157
```
157158

158159
</details>
@@ -170,7 +171,7 @@ pip install paddle2onnx
170171
- Convert model:
171172

172173
```shell
173-
paddle2onnx --model_dir output_inference/picodet_s_320_coco/ \
174+
paddle2onnx --model_dir output_inference/picodet_s_320_coco_lcnet/ \
174175
--model_filename model.pdmodel \
175176
--params_filename model.pdiparams \
176177
--opset_version 11 \
@@ -245,7 +246,7 @@ pip install paddleslim==2.2.1
245246
Configure the quant config and start training:
246247

247248
```shell
248-
python tools/train.py -c configs/picodet/picodet_s_320_coco.yml \
249+
python tools/train.py -c configs/picodet/picodet_s_320_coco_lcnet.yml \
249250
--slim_config configs/slim/quant/picodet_s_quant.yml --eval
250251
```
251252

@@ -259,7 +260,7 @@ python tools/train.py -c configs/picodet/picodet_s_320_coco.yml \
259260
Configure the post quant config and start calibrate model:
260261

261262
```shell
262-
python tools/post_quant.py -c configs/picodet/picodet_s_320_coco.yml \
263+
python tools/post_quant.py -c configs/picodet/picodet_s_320_coco_lcnet.yml \
263264
--slim_config configs/slim/post_quant/picodet_s_ptq.yml
264265
```
265266

@@ -296,7 +297,7 @@ Please reduce the `batch_size` of `TrainReader` in config.
296297

297298
Please reset `pretrain_weights` in config, which trained on coco. Such as:
298299
```yaml
299-
pretrain_weights: https://paddledet.bj.bcebos.com/models/picodetv2/picodet_l_640_coco.pdparams
300+
pretrain_weights: https://paddledet.bj.bcebos.com/models/picodet_l_640_coco_lcnet.pdparams
300301
```
301302
302303
</details>

0 commit comments

Comments
 (0)