{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. Introduction\n",
    "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.\n",
    "\n",
    "The PP-PicoDet model has the following characteristics:\n",
    "- 🌟 Higher mAP: the **first** object detectors that surpass mAP(0.5:0.95) **30+** within 1M parameters when the input size is 416.\n",
    "- 🚀 Faster latency: 150FPS on mobile ARM CPU.\n",
    "- 😊 Deploy friendly: support PaddleLite/MNN/NCNN/OpenVINO and provide C++/Python/Android implementation.\n",
    "- 😍 Advanced algorithm: use the most advanced algorithms and offer innovation, such as ESNet, CSP-PAN, SimOTA with VFL, etc.\n",
    "\n",
    "For more details, please refer to [official documentation](https://github.com/PaddlePaddle/PaddleDetection/blob/release/2.5/configs/picodet/README_en.md)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. Model Effects\n",
    "The accuracy and speed comparison of PP-Picodet and other lightweight models is shown below:\n",
    "<div align=\"center\">\n",
    "  <img src=\"https://raw.githubusercontent.com/PaddlePaddle/PaddleDetection/release/2.5/docs/images/picodet_map.png\" width=500 />\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3. How to use the model\n",
    "Clone PaddleDetection firstly and put the COCO-style dataset in `dataset/coco`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# clone PaddleDetection\n",
    "%mkdir -p ~/work\n",
    "%cd ~/work/\n",
    "!git clone https://github.com/PaddlePaddle/PaddleDetection.git\n",
    "\n",
    "# Other Dependencies\n",
    "%cd PaddleDetection\n",
    "%mkdir -p demo_input demo_output\n",
    "!pip install -r requirements.txt"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.1 Training\n",
    "Training PP-Picodet with following command"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# training with single GPU\n",
    "!CUDA_VISIBLE_DEVICES=0 python tools/train.py -c configs/picodet/picodet_s_320_coco_lcnet.yml --eval\n",
    "\n",
    "# training with mutiple GPUs\n",
    "!CUDA_VISIBLE_DEVICES=0,1,2,3 python -m paddle.distributed.launch --gpus 0,1,2,3 tools/train.py -c configs/picodet/picodet_s_320_coco_lcnet.yml --eval"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Notes:**\n",
    "- All models of PicoDet are trained by 4 GPUs. If the number of GPUs is changed, the learning rate `base_lr` needs to be scaled linearly."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.2 Deployment\n",
    "\n",
    "Run the following command to quickly deploy the model using PaddleInference"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# export the model\n",
    "!python tools/export_model.py -c configs/picodet/picodet_s_320_coco_lcnet.yml -o weights=https://paddledet.bj.bcebos.com/models/picodet_s_320_coco_lcnet.pdparams\n",
    "\n",
    "# inference\n",
    "!wget -P demo_input -N https://paddledet.bj.bcebos.com/modelcenter/images/General/000000014439.jpg\n",
    "!python deploy/python/infer.py --model_dir=output_inference/picodet_s_320_coco_lcnet --image_file=demo_input/000000014439.jpg --device=GPU --output_dir=demo_output"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "PP-Picodet supports multiple deployment methods, please refer to [PP-Picodet deployment](https://github.com/PaddlePaddle/PaddleDetection/blob/release/2.5/configs/picodet/README_en.md#deployment) for details."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4. Model principle\n",
    "The overall structure of PP-Picodet is shown below:\n",
    "<div align=\"center\">\n",
    "  <img src=\"https://bj.bcebos.com/v1/paddledet/modelcenter/images/PP-Picodet-arch.png\" width=70% />\n",
    "</div>\n",
    "PP-Picodet is composed of following methods:\n",
    "- Enhanced ShuffleNet-ESNet\n",
    "- CSP-PAN\n",
    "- SimOTA label assignment\n",
    "\n",
    "For more details, please refer to our technical report: https://arxiv.org/abs/2111.00902"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5. Attention\n",
    "**All commands run on AI Studio's `jupyter` by default. If running on a terminal, remove the % or ! at the beginning of the command.**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 6. Related papers and citations\n",
    "```\n",
    "@article{yu2021pp,\n",
    "  title={PP-PicoDet: A Better Real-Time Object Detector on Mobile Devices},\n",
    "  author={Yu, Guanghua and Chang, Qinyao and Lv, Wenyu and Xu, Chang and Cui, Cheng and Ji, Wei and Dang, Qingqing and Deng, Kaipeng and Wang, Guanzhong and Du, Yuning and others},\n",
    "  journal={arXiv preprint arXiv:2111.00902},\n",
    "  year={2021}\n",
    "}\n",
    "```"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.10.6 64-bit",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "name": "python",
   "version": "3.10.6"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "31f2aee4e71d21fbe5cf8b01ff0e069b9275f58929596ceb00d14d90e3e16cd6"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}