update PicoDet docs (#4333)

* update PicoDet docs

update PicoDet docs (#4333)
* update PicoDet docs
3ee7bde2 · Guanghua Yu · GitHub · 488a6838 · 3ee7bde2 · 3ee7bde2
7 changed file
--- a/configs/picodet/README.md
+++ b/configs/picodet/README.md
@@ -5,44 +5,146 @@
 We developed a series of lightweight models, which named `PicoDet`. Because of its excellent performance, it is very suitable for deployment on mobile or CPU.
- 🌟 Higher mAP:  The **first** model which within 1M parameter with mAP reaching 30+.
+- 🌟 Higher mAP: the **first** object detectors that surpass mAP(0.5:0.95) **30+** within 1M parameters when the input size is 416.
 - 🚀 Faster latency: 114FPS on mobile ARM CPU.
 - 😊 Deploy friendly: support PaddleLite/MNN/NCNN/OpenVINO and provide C++/Python/Android implementation.
 - 😍 Advanced algorithm: use the most advanced algorithms and innovate, such as ESNet, CSP-PAN, SimOTA with VFL, etc.
+### Comming soon
+- [ ] More series of model, such as smaller or larger model.
+- [ ] Pretrained models for more scenarios.
+- [ ] More features in need.
 ## Requirements
 - PaddlePaddle >= 2.1.2
 - PaddleSlim >= 2.1.1
-## Comming soon
- [ ] More series of model, such as Smaller or larger model.
- [ ] Pretrained models for more scenarios.
- [ ] More features in need.
 ## Model Zoo
-| Model                  | Input size | lr schedule | Box AP(0.5:0.95) | Box AP(0.5) | FLOPS | Model Size | Inference Time |                           download                          | config |
+| Model                  | Input size | mAP<sup>val<br>0.5:0.95 | mAP<sup>val<br>0.5 | FLOPS<br><sup>(G) | Params<br><sup>(M) | Latency<br><sup>(ms) |                           download                          | config |
-| :------------------------ | :-------: | :-------: | :------: | :---: | :---: | :---: | :------------:  | :-------------------------------------------------: | :-----: |
+| :------------------------ | :-------:  | :------: | :---: | :---: | :---: | :------------:  | :-------------------------------------------------: | :-----: |
-| PicoDet-S    | 320*320    |   300e      |   27.1     | 41.4 | -- | 3.9M | -- | [model](https://paddledet.bj.bcebos.com/models/picodet_s_320_coco.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/train_picodet_s_320_coco.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_s_320_coco.yml) |
+| PicoDet-S    | 320*320    |   27.1     | 41.4 | -- | 3.9M | -- | [model](https://paddledet.bj.bcebos.com/models/picodet_s_320_coco.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/train_picodet_s_320_coco.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_s_320_coco.yml) |
-| PicoDet-S    | 416*416    |   300e      |   30.6     | 45.5 | -- | 3.9M | -- | [model](https://paddledet.bj.bcebos.com/models/picodet_s_416_coco.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/train_picodet_s_416_coco.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_s_416_coco.yml) |
+| PicoDet-M    | 320*320   |  30.9     | 45.7 |  -- | 8.4M | -- | [model](https://paddledet.bj.bcebos.com/models/picodet_m_320_coco.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/train_picodet_m_320_coco.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_m_320_coco.yml) |
-| PicoDet-M    | 320*320   |   300e      |  -     | 41.2 |  -- | 8.4M | -- | [model](https://paddledet.bj.bcebos.com/models/picodet_m_320_coco.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/train_picodet_m_320_coco.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_m_320_coco.yml) |
+| PicoDet-L    | 320*320   |  32.6     | 47.9 |  -- | 13M | -- | [model](https://paddledet.bj.bcebos.com/models/picodet_l_320_coco.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/train_picodet_l_320_coco.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_l_320_coco.yml) |
-| PicoDet-M    | 416*416   |   300e      |   34.3     | 49.8 |  -- | 8.4M | -- | [model](https://paddledet.bj.bcebos.com/models/picodet_m_416_coco.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/train_picodet_m_416_coco.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_m_416_coco.yml) |
+| PicoDet-S    | 416*416    |   30.6     | 45.5 | -- | 3.9M | -- | [model](https://paddledet.bj.bcebos.com/models/picodet_s_416_coco.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/train_picodet_s_416_coco.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_s_416_coco.yml) |
+| PicoDet-M    | 416*416    |   34.3     | 49.8 |  -- | 8.4M | -- | [model](https://paddledet.bj.bcebos.com/models/picodet_m_416_coco.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/train_picodet_m_416_coco.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_m_416_coco.yml) |
+| PicoDet-L    | 416*416   |  -    |   -  |  -- | 13M | -- | [model](https://paddledet.bj.bcebos.com/models/picodet_l_416_coco.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/train_picodet_l_416_coco.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_l_416_coco.yml) |
+| PicoDet-L    | 640*640   |  -    |  -   |  -- | 13M | -- | [model](https://paddledet.bj.bcebos.com/models/picodet_l_640_coco.pdparams) &#124; [log](https://paddledet.bj.bcebos.com/logs/train_picodet_l_640_coco.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/picodet_l_640_coco.yml) |
 **Notes:**
- PicoDet inference speed is tested on Kirin 980 with 4 threads by arm8 and with FP16.
+- PicoDet inference speed is tested on Snapdragon 888(4xA78+4xA55) with 4 threads by arm8 and with FP16.
 - PicoDet is trained on COCO train2017 dataset and evaluated on val2017.
- PicoDet used 4 GPUs for training and mini-batch size as 128 or 96 on each GPU.
+- PicoDet used 4 or 8 GPUs for training.
+## Deployment
+### Export and Convert model
+<details>
+<summary>1. Export model</summary>
+```shell
+cd PaddleDetection
+python tools/export_model.py -c configs/picodet/picodet_s_320_coco.yml \
+              -o weights=https://paddledet.bj.bcebos.com/models/picodet_s_320_coco.pdparams --output_dir=inference_model
+```
+</details>
+<details>
+<summary>2. Convert to PaddleLite</summary>
+- Install Paddlelite>=2.10.rc:
+```shell
+pip install paddlelite
+```
+- Convert model:
+```shell
+# FP32
+paddle_lite_opt --model_dir=inference_model/picodet_s_320_coco --valid_targets=arm --optimize_out=picodet_s_320_coco_fp32
+# FP16
+paddle_lite_opt --model_dir=inference_model/picodet_s_320_coco --valid_targets=arm --optimize_out=picodet_s_320_coco_fp16 --enable_fp16=true
+```
+</details>
+<details>
+<summary>3. Convert to ONNX</summary>
+- Install [Paddle2ONNX](https://github.com/PaddlePaddle/Paddle2ONNX) >= 0.7 and ONNX > 1.10.1, for details, please refer to [Tutorials of Export ONNX Model](../../deploy/EXPORT_ONNX_MODEL.md)
+```shell
+pip install onnx
+pip install paddle2onnx
+```
+- Convert model:
+```shell
+paddle2onnx --model_dir output_inference/picodet_s_320_coco/ \
+            --model_filename model.pdmodel  \
+            --params_filename model.pdiparams \
+            --opset_version 11 \
+            --save_file picodet_s_320_coco.onnx
+```
+- Simplify ONNX model: use onnx-simplifier to simplify onnx model.
+  - Install onnx-simplifier >= 0.3.6:
+  ```shell
+  pip install onnx-simplifier
+  ```
+  - simplify onnx model:
+  ```shell
+  python -m onnxsim picodet_s_320_coco.onnx picodet_s_processed.onnx
+  ```
+</details>
+### Deploy
+- PaddleInference demo [Python](../../deploy/python) & [C++](../../deploy/cpp)
+- [PaddleLite C++ demo](../../deploy/lite)
+- [NCNN C++ demo]()
+- [MNN C++ demo]()
+- [OpenVINO C++ demo]()
+- [Android demo]()
+## Slim
+### quantization
+<details>
+<summary>Quant aware</summary>
+Configure the quant config and start training:
+```shell
+python tools/train.py -c configs/picodet/picodet_s_320_coco.yml \
+          --slim_config configs/slim/quant/picodet_s_quant.yml --eval
+```
+</details>
+<details>
+<summary>Post quant</summary>
+Configure the post quant config and start calibrate model:
+```shell
+python tools/posy_quant.py -c configs/picodet/picodet_s_320_coco.yml \
+          --slim_config configs/slim/posy_quant/picodet_s_quant.yml
+```
+</details>
-## Citations
+## Cite PiocDet
+If you use PiocDet in your research, please cite our work by using the following BibTeX entry:
 ```
-@article{li2020generalized,
+comming soon
-  title={Generalized Focal Loss: Learning Qualified and Distributed Bounding Boxes for Dense Object Detection},
-  author={Li, Xiang and Wang, Wenhai and Wu, Lijun and Chen, Shuo and Hu, Xiaolin and Li, Jun and Tang, Jinhui and Yang, Jian},
-  journal={arXiv preprint arXiv:2006.04388},
-  year={2020}
-}
 ```
--- a/configs/picodet/picodet_l_640_coco.yml
+++ b/configs/picodet/picodet_l_640_coco.yml
@@ -31,7 +31,7 @@ PicoHead:
  feat_in_chan: 128
 TrainReader:
-  batch_size: 48
+  batch_size: 32
 LearningRate:
  base_lr: 0.3

--- a/configs/slim/post_quant/picodet_s_ptq.yml
+++ b/configs/slim/post_quant/picodet_s_ptq.yml
+weights:  https://paddledet.bj.bcebos.com/models/picodet_s_320_coco.pdparams
+slim: PTQ
+PTQ:
+  ptq_config: {
+      'activation_quantizer': 'HistQuantizer',
+      'upsample_bins': 127,
+      'hist_percent': 0.999}
+  quant_batch_num: 10
+  fuse: True
--- a/configs/slim/quant/picodet_s_shufflenetv2_quant.yml
+++ b/configs/slim/quant/picodet_s_shufflenetv2_quant.yml
-pretrain_weights: https://paddledet.bj.bcebos.com/models/picodet_s_shufflenetv2_320_coco.pdparams
+pretrain_weights: https://paddledet.bj.bcebos.com/models/picodet_s_320_coco.pdparams
 slim: QAT
 QAT:
@@ -9,15 +9,15 @@ QAT:
    'quantizable_layer_type': ['Conv2D', 'Linear']}
  print_model: False
-epoch: 80
+epoch: 50
 LearningRate:
  base_lr: 0.001
  schedulers:
  - !PiecewiseDecay
    gamma: 0.1
    milestones:
-    - 60
+    - 30
-    - 70
+    - 40
  - !LinearWarmup
    start_factor: 0.
    steps: 100

--- a/deploy/EXPORT_ONNX_MODEL.md
+++ b/deploy/EXPORT_ONNX_MODEL.md
@@ -11,6 +11,7 @@ PaddleDetection模型支持保存为ONNX格式，目前测试支持的列表如
 | PAFNet | 11 |- |
 | TTFNet | 11 |-|
 | SSD | 11 |仅支持batch=1推理 |
+| PicoDet | 11 |仅支持batch=1推理 |
 保存ONNX的功能由[Paddle2ONNX](https://github.com/PaddlePaddle/Paddle2ONNX)提供，如在转换中有相关问题反馈，可在Paddle2ONNX的Github项目中通过[ISSUE](https://github.com/PaddlePaddle/Paddle2ONNX/issues)与工程师交流。

--- a/ppdet/modeling/assigners/simota_assigner.py
+++ b/ppdet/modeling/assigners/simota_assigner.py
@@ -196,7 +196,6 @@ class SimOTAAssigner(object):
        num_valid = valid_decoded_bbox.shape[0]
        pairwise_ious = batch_bbox_overlaps(valid_decoded_bbox, gt_bboxes)
-        iou_cost = -paddle.log(pairwise_ious + eps)
        if self.use_vfl:
            gt_vfl_labels = gt_labels.squeeze(-1).unsqueeze(0).tile(
                [num_valid, 1]).reshape([-1])
@@ -216,6 +215,7 @@ class SimOTAAssigner(object):
                paddle.logical_not(is_in_boxes_and_center).cast('float32') * INF
            )
        else:
+            iou_cost = -paddle.log(pairwise_ious + eps)
            gt_onehot_label = (F.one_hot(
                gt_labels.squeeze(-1).cast(paddle.int64),
                pred_scores.shape[-1]).cast('float32').unsqueeze(0).tile(

--- a/ppdet/modeling/losses/varifocal_loss.py
+++ b/ppdet/modeling/losses/varifocal_loss.py
@@ -34,9 +34,9 @@ def varifocal_loss(pred,
    """`Varifocal Loss <https://arxiv.org/abs/2008.13367>`_
    Args:
-        pred (torch.Tensor): The prediction with shape (N, C), C is the
+        pred (Tensor): The prediction with shape (N, C), C is the
            number of classes
-        target (torch.Tensor): The learning target of the iou-aware
+        target (Tensor): The learning target of the iou-aware
            classification score with shape (N, C), C is the number of classes.
        alpha (float, optional): A balance factor for the negative part of
            Varifocal Loss, which is different from the alpha of Focal Loss.
@@ -108,27 +108,18 @@ class VarifocalLoss(nn.Layer):
        self.reduction = reduction
        self.loss_weight = loss_weight
-    def forward(self,
+    def forward(self, pred, target, weight=None, avg_factor=None):
-                pred,
-                target,
-                weight=None,
-                avg_factor=None,
-                reduction_override=None):
        """Forward function.
        Args:
-            pred (torch.Tensor): The prediction.
+            pred (Tensor): The prediction.
-            target (torch.Tensor): The learning target of the prediction.
+            target (Tensor): The learning target of the prediction.
-            weight (torch.Tensor, optional): The weight of loss for each
+            weight (Tensor, optional): The weight of loss for each
                prediction. Defaults to None.
            avg_factor (int, optional): Average factor that is used to average
                the loss. Defaults to None.
-            reduction_override (str, optional): The reduction method used to
-                override the original reduction method of the loss.
-                Options are "none", "mean" and "sum".
        Returns:
-            torch.Tensor: The calculated loss
+            Tensor: The calculated loss
        """
        loss = self.loss_weight * varifocal_loss(
            pred,