# PP-PicoDet ![](../../docs/images/picedet_demo.jpeg) ## Introduction 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). - 🌟 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: 150FPS 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 offer innovation, 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 ## Benchmark | Model | Input size | mAPval
0.5:0.95 | mAPval
0.5 | Params
(M) | FLOPS
(G) | Latency[NCNN](#latency)
(ms) | Latency[Lite](#latency)
(ms) | download | config | | :-------- | :--------: | :---------------------: | :----------------: | :----------------: | :---------------: | :-----------------------------: | :-----------------------------: | :---------------------------------------------------------------------------------------------------------------------------------------------------------: | :------------------------------------------------------------------------------------------------------------ | | PicoDet-S | 320*320 | 27.1 | 41.4 | 0.99 | 0.73 | 8.13 | **6.65** | [model](https://paddledet.bj.bcebos.com/models/picodet_s_320_coco.pdparams) | [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 | 30.6 | 45.5 | 0.99 | 1.24 | 12.37 | **9.82** | [model](https://paddledet.bj.bcebos.com/models/picodet_s_416_coco.pdparams) | [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 | 2.15 | 1.48 | 11.27 | **9.61** | [model](https://paddledet.bj.bcebos.com/models/picodet_m_320_coco.pdparams) | [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 | 416*416 | 34.3 | 49.8 | 2.15 | 2.50 | 17.39 | **15.88** | [model](https://paddledet.bj.bcebos.com/models/picodet_m_416_coco.pdparams) | [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 | 320*320 | 32.9 | 48.2 | 3.30 | 2.23 | 15.26 | **13.42** | [model](https://paddledet.bj.bcebos.com/models/picodet_l_320_coco.pdparams) | [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-L | 416*416 | 36.6 | 52.5 | 3.30 | 3.76 | 23.36 | **21.85** | [model](https://paddledet.bj.bcebos.com/models/picodet_l_416_coco.pdparams) | [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 | 40.9 | 57.6 | 3.30 | 8.91 | 54.11 | **50.55** | [model](https://paddledet.bj.bcebos.com/models/picodet_l_640_coco.pdparams) | [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) | #### More Configs | Model | Input size | mAPval
0.5:0.95 | mAPval
0.5 | Params
(M) | FLOPS
(G) | Latency[NCNN](#latency)
(ms) | Latency[Lite](#latency)
(ms) | download | config | | :--------------------------- | :--------: | :---------------------: | :----------------: | :----------------: | :---------------: | :-----------------------------: | :-----------------------------: | :-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: | :------------------------------------------------------------------------------------------------------------------------------------------- | | PicoDet-Shufflenetv2 1x | 416*416 | 30.0 | 44.6 | 1.17 | 1.53 | 15.06 | **10.63** | [model](https://paddledet.bj.bcebos.com/models/picodet_shufflenetv2_1x_416_coco.pdparams) | [log](https://paddledet.bj.bcebos.com/logs/train_picodet_shufflenetv2_1x_416_coco.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/more_config/picodet_shufflenetv2_1x_416_coco.yml) | | PicoDet-MobileNetv3-large 1x | 416*416 | 35.6 | 52.0 | 3.55 | 2.80 | 20.71 | **17.88** | [model](https://paddledet.bj.bcebos.com/models/picodet_mobilenetv3_large_1x_416_coco.pdparams) | [log](https://paddledet.bj.bcebos.com/logs/train_picodet_mobilenetv3_large_1x_416_coco.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/more_config/picodet_mobilenetv3_large_1x_416_coco.yml) | | PicoDet-LCNet 1.5x | 416*416 | 36.3 | 52.2 | 3.10 | 3.85 | 21.29 | **20.8** | [model](https://paddledet.bj.bcebos.com/models/picodet_lcnet_1_5x_416_coco.pdparams) | [log](https://paddledet.bj.bcebos.com/logs/train_picodet_lcnet_1_5x_416_coco.log) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/develop/configs/picodet/more_config/picodet_lcnet_1_5x_416_coco.yml) |
Table Notes: - Latency: All our models test on `Qualcomm Snapdragon 865(4xA77+4xA55)` with 4 threads by arm8 and with FP16. In the above table, test latency on [NCNN](https://github.com/Tencent/ncnn) and `Lite`->[Paddle-Lite](https://github.com/PaddlePaddle/Paddle-Lite). And testing latency with code: [MobileDetBenchmark](https://github.com/JiweiMaster/MobileDetBenchmark). - PicoDet is trained on COCO train2017 dataset and evaluated on COCO val2017. - PicoDet used 4 or 8 GPUs for training and all checkpoints are trained with default settings and hyperparameters.
#### Benchmark of Other Models | Model | Input size | mAPval
0.5:0.95 | mAPval
0.5 | Params
(M) | FLOPS
(G) | Latency[NCNN](#latency)
(ms) | | :-------- | :--------: | :---------------------: | :----------------: | :----------------: | :---------------: | :-----------------------------: | | YOLOv3-Tiny | 416*416 | 16.6 | 33.1 | 8.86 | 5.62 | 25.42 | | YOLOv4-Tiny | 416*416 | 21.7 | 40.2 | 6.06 | 6.96 | 23.69 | | PP-YOLO-Tiny | 320*320 | 20.6 | - | 1.08 | 0.58 | 6.75 | | PP-YOLO-Tiny | 416*416 | 22.7 | - | 1.08 | 1.02 | 10.48 | | Nanodet-M | 320*320 | 20.6 | - | 0.95 | 0.72 | 8.71 | | Nanodet-M | 416*416 | 23.5 | - | 0.95 | 1.2 | 13.35 | | Nanodet-M 1.5x | 416*416 | 26.8 | - | 2.08 | 2.42 | 15.83 | | YOLOX-Nano | 416*416 | 25.8 | - | 0.91 | 1.08 | 19.23 | | YOLOX-Tiny | 416*416 | 32.8 | - | 5.06 | 6.45 | 32.77 | | YOLOv5n | 640*640 | 28.4 | 46.0 | 1.9 | 4.5 | 40.35 | | YOLOv5s | 640*640 | 37.2 | 56.0 | 7.2 | 16.5 | 78.05 | ## Deployment ### Export and Convert Model
1. Export model (click to expand) ```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 ```
2. Convert to PaddleLite (click to expand) - 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 ```
3. Convert to ONNX (click to expand) - 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 ```
### Deploy - PaddleInference demo [Python](../../deploy/python) & [C++](../../deploy/cpp) - [PaddleLite C++ demo](../../deploy/lite) - [NCNN C++/Python demo](../../deploy/third_engine/demo_ncnn) - [MNN C++/Python demo](../../deploy/third_engine/demo_mnn) - [OpenVINO C++/Python demo](../../deploy/third_engine/demo_openvino) - [Android demo](https://github.com/JiweiMaster/PP-PicoDet-Android-Demo) Android demo visualization:
## Quantization
Requirements: - PaddlePaddle >= 2.2.0rc0 - PaddleSlim >= 2.2.0rc0 **Install:** ```shell pip install paddleslim==2.2.0rc0 ```
Quant aware (click to expand) 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 ```
Post quant (click to expand) Configure the post quant config and start calibrate model: ```shell python tools/post_quant.py -c configs/picodet/picodet_s_320_coco.yml \ --slim_config configs/slim/post_quant/picodet_s_ptq.yml ``` - Notes: Now the accuracy of post quant is abnormal and this problem is being solved.
## FAQ
Out of memory error. Please reduce the `batch_size` of `TrainReader` in config.
## Cite PP-PiocDet If you use PiocDet in your research, please cite our work by using the following BibTeX entry: ``` @misc{yu2021pppicodet, title={PP-PicoDet: A Better Real-Time Object Detector on Mobile Devices}, author={Guanghua Yu and Qinyao Chang and Wenyu Lv and Chang Xu and Cheng Cui and Wei Ji and Qingqing Dang and Kaipeng Deng and Guanzhong Wang and Yuning Du and Baohua Lai and Qiwen Liu and Xiaoguang Hu and Dianhai Yu and Yanjun Ma}, year={2021}, eprint={2111.00902}, archivePrefix={arXiv}, primaryClass={cs.CV} } ```