简体中文 | [English](README_en.md) # Semi-Supervised Object Detection (SSOD) 半监督目标检测 ## 内容 - [简介](#简介) - [模型库](#模型库) - [Baseline](#Baseline) - [DenseTeacher](#DenseTeacher) - [半监督数据集准备](#半监督数据集准备) - [半监督检测配置](#半监督检测配置) - [训练集配置](#训练集配置) - [预训练配置](#预训练配置) - [全局配置](#全局配置) - [模型配置](#模型配置) - [数据增强配置](#数据增强配置) - [其他配置](#其他配置) - [使用说明](#使用说明) - [训练](#训练) - [评估](#评估) - [预测](#预测) - [部署](#部署) - [引用](#引用) ## 简介 半监督目标检测(SSOD)是**同时使用有标注数据和无标注数据**进行训练的目标检测,既可以极大地节省标注成本,也可以充分利用无标注数据进一步提高检测精度。PaddleDetection团队复现了[DenseTeacher](denseteacher)半监督检测算法,用户可以下载使用。 ## 模型库 ### [Baseline](baseline) **纯监督数据**模型的训练和模型库,请参照[Baseline](baseline); ### [DenseTeacher](denseteacher) | 模型 | 基础检测器 | 监督数据比例 | Sup mAPval
0.5:0.95 | Semi mAPval
0.5:0.95 | Semi Epochs (Iters) | 模型下载 | 配置文件 | | :------------: | :---------------------: | :-----------: | :-------------------------: |:---------------------------: |:--------------------: | :-------: |:---------: | | DenseTeacher | [FCOS ResNet50-FPN](./baseline/fcos_r50_fpn_2x_coco_sup005.yml) | 5% | 21.3 | 30.6 | 240 (87120) | [download](https://paddledet.bj.bcebos.com/models/denseteacher_fcos_r50_fpn_coco_semi005.pdparams) | [config](denseteacher/denseteacher_fcos_r50_fpn_coco_semi005.yml) | | DenseTeacher | [FCOS ResNet50-FPN](./baseline/fcos_r50_fpn_2x_coco_sup010.yml) | 10%| 26.3 | 35.1 | 240 (174240)| [download](https://paddledet.bj.bcebos.com/models/denseteacher_fcos_r50_fpn_coco_semi010.pdparams) | [config](denseteacher/denseteacher_fcos_r50_fpn_coco_semi010.yml) | | DenseTeacher(LSJ)| [FCOS ResNet50-FPN](./baseline/fcos_r50_fpn_2x_coco_sup010.yml) |10%| 26.3 | 37.1 | 240 (174240)| [download](https://paddledet.bj.bcebos.com/models/denseteacher_fcos_r50_fpn_coco_semi010_lsj.pdparams) | [config](denseteacher/denseteacher_fcos_r50_fpn_coco_semi010_lsj.yml) | | DenseTeacher | [FCOS ResNet50-FPN](./../fcos/fcos_r50_fpn_iou_multiscale_2x_coco.ymll) |full| 42.6 | - | 36 (263844)| [download](https://paddledet.bj.bcebos.com/models/denseteacher_fcos_r50_fpn_coco_full.pdparams) | [config](denseteacher/denseteacher_fcos_r50_fpn_coco_full.yml) | ## 半监督数据集准备 半监督目标检测**同时需要有标注数据和无标注数据**,且无标注数据量一般**远多于有标注数据量**。 对于COCO数据集一般有两种常规设置: (1)抽取部分比例的原始训练集`train2017`作为标注数据和无标注数据; 从`train2017`中按固定百分比(1%、2%、5%、10%等)抽取,由于抽取方法会对半监督训练的结果影响较大,所以采用五折交叉验证来评估。运行数据集划分制作的脚本如下: ```bash python tools/gen_semi_coco.py ``` 会按照 1%、2%、5%、10% 的监督数据比例来划分`train2017`全集,为了交叉验证每一种划分会随机重复5次,生成的半监督标注文件如下: - 标注数据集标注:`instances_train2017.{fold}@{percent}.json` - 无标注数据集标注:`instances_train2017.{fold}@{percent}-unlabeled.json` 其中,`fold` 表示交叉验证,`percent` 表示有标注数据的百分比。 (2)使用全量原始训练集`train2017`作为有标注数据 和 全量原始无标签图片集`unlabeled2017`作为无标注数据; ### 下载链接 PaddleDetection团队提供了COCO数据集全部的标注文件,请下载并解压存放至对应目录: ```shell # 下载COCO全量数据集图片和标注 # 包括 train2017, val2017, annotations wget https://bj.bcebos.com/v1/paddledet/data/coco.tar # 下载PaddleDetection团队整理的COCO部分比例数据的标注文件 wget https://bj.bcebos.com/v1/paddledet/data/coco/semi_annotations.zip # unlabeled2017是可选,如果不需要训‘full’则无需下载 # 下载COCO全量 unlabeled 无标注数据集 wget https://bj.bcebos.com/v1/paddledet/data/coco/unlabeled2017.zip wget https://bj.bcebos.com/v1/paddledet/data/coco/image_info_unlabeled2017.zip # 下载转换完的 unlabeled2017 无标注json文件 wget https://bj.bcebos.com/v1/paddledet/data/coco/instances_unlabeled2017.zip ``` 如果需要用到COCO全量unlabeled无标注数据集,需要将原版的`image_info_unlabeled2017.json`进行格式转换,运行以下代码:
COCO unlabeled 标注转换代码: ```python import json anns_train = json.load(open('annotations/instances_train2017.json', 'r')) anns_unlabeled = json.load(open('annotations/image_info_unlabeled2017.json', 'r')) unlabeled_json = { 'images': anns_unlabeled['images'], 'annotations': [], 'categories': anns_train['categories'], } path = 'annotations/instances_unlabeled2017.json' with open(path, 'w') as f: json.dump(unlabeled_json, f) ```
解压后的数据集目录如下: ``` PaddleDetection ├── dataset │ ├── coco │ │ ├── annotations │ │ │ ├── instances_train2017.json │ │ │ ├── instances_unlabeled2017.json │ │ │ ├── instances_val2017.json │ │ ├── semi_annotations │ │ │ ├── instances_train2017.1@1.json │ │ │ ├── instances_train2017.1@1-unlabeled.json │ │ │ ├── instances_train2017.1@2.json │ │ │ ├── instances_train2017.1@2-unlabeled.json │ │ │ ├── instances_train2017.1@5.json │ │ │ ├── instances_train2017.1@5-unlabeled.json │ │ │ ├── instances_train2017.1@10.json │ │ │ ├── instances_train2017.1@10-unlabeled.json │ │ ├── train2017 │ │ ├── unlabeled2017 │ │ ├── val2017 ```
## 半监督检测配置 配置半监督检测,需要基于选用的**基础检测器**的配置文件,如: ```python _BASE_: [ '../../fcos/fcos_r50_fpn_iou_multiscale_2x_coco.yml', '../_base_/coco_detection_percent_10.yml', ] log_iter: 50 snapshot_epoch: 5 epochs: &epochs 240 weights: output/denseteacher_fcos_r50_fpn_coco_semi010/model_final ``` 并依次做出如下几点改动: ### 训练集配置 首先可以直接引用已经配置好的半监督训练集,如: ```python _BASE_: [ '../_base_/coco_detection_percent_10.yml', ] ``` 具体来看,构建半监督数据集,需要同时配置监督数据集`TrainDataset`和无监督数据集`UnsupTrainDataset`的路径,**注意必须选用`SemiCOCODataSet`类而不是`COCODataSet`类**,如以下所示: **COCO-train2017部分比例数据集**: ```python # partial labeled COCO, use `SemiCOCODataSet` rather than `COCODataSet` TrainDataset: !SemiCOCODataSet image_dir: train2017 anno_path: semi_annotations/instances_train2017.1@1.json dataset_dir: dataset/coco data_fields: ['image', 'gt_bbox', 'gt_class', 'is_crowd'] # partial unlabeled COCO, use `SemiCOCODataSet` rather than `COCODataSet` UnsupTrainDataset: !SemiCOCODataSet image_dir: train2017 anno_path: semi_annotations/instances_train2017.1@1-unlabeled.json dataset_dir: dataset/coco data_fields: ['image'] supervised: False ``` 或者 **COCO-train2017 full全量数据集**: ```python # full labeled COCO, use `SemiCOCODataSet` rather than `COCODataSet` TrainDataset: !SemiCOCODataSet image_dir: train2017 anno_path: annotations/instances_train2017.json dataset_dir: dataset/coco data_fields: ['image', 'gt_bbox', 'gt_class', 'is_crowd'] # full unlabeled COCO, use `SemiCOCODataSet` rather than `COCODataSet` UnsupTrainDataset: !SemiCOCODataSet image_dir: unlabeled2017 anno_path: annotations/instances_unlabeled2017.json dataset_dir: dataset/coco data_fields: ['image'] supervised: False ``` 验证集`EvalDataset`和测试集`TestDataset`的配置**不需要更改**,且还是采用`COCODataSet`类。 ### 预训练配置 ```python ### pretrain and warmup config, choose one and coment another pretrain_weights: https://paddledet.bj.bcebos.com/models/pretrained/ResNet50_cos_pretrained.pdparams semi_start_iters: 5000 ema_start_iters: 3000 use_warmup: &use_warmup True ``` ### 全局配置 需要在配置文件中添加如下全局配置: ```python ### global config use_simple_ema: True ema_decay: 0.9996 ssod_method: DenseTeacher DenseTeacher: train_cfg: sup_weight: 1.0 unsup_weight: 1.0 loss_weight: {distill_loss_cls: 4.0, distill_loss_box: 1.0, distill_loss_quality: 1.0} concat_sup_data: True suppress: linear ratio: 0.01 gamma: 2.0 test_cfg: inference_on: teacher ``` ### 模型配置 如果没有特殊改动,则直接继承自基础检测器里的模型配置。 以 `DenseTeacher` 为例,选择 `fcos_r50_fpn_iou_multiscale_2x_coco.yml` 作为**基础检测器**进行半监督训练,**teacher网络的结构和student网络的结构均为基础检测器的结构,且结构相同**。 ```python _BASE_: [ '../../fcos/fcos_r50_fpn_iou_multiscale_2x_coco.yml', ] ``` ### 数据增强配置 构建半监督训练集的Reader,需要在原先`TrainReader`的基础上,新增加`weak_aug`,`strong_aug`,`sup_batch_transforms`和`unsup_batch_transforms`,并且需要注意: - **如果有`NormalizeImage`,需要单独从`sample_transforms`中抽出来放在`weak_aug`和`strong_aug`中; - `sample_transforms`为**公用的基础数据增强**; - 完整的弱数据增强为``sample_transforms + weak_aug`,完整的强数据增强为`sample_transforms + strong_aug`; 如以下所示: 原纯监督模型的`TrainReader`: ```python TrainReader: sample_transforms: - Decode: {} - RandomResize: {target_size: [[640, 1333], [672, 1333], [704, 1333], [736, 1333], [768, 1333], [800, 1333]], keep_ratio: True, interp: 1} - RandomFlip: {} - NormalizeImage: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True} batch_transforms: - Permute: {} - PadBatch: {pad_to_stride: 32} - Gt2FCOSTarget: object_sizes_boundary: [64, 128, 256, 512] center_sampling_radius: 1.5 downsample_ratios: [8, 16, 32, 64, 128] norm_reg_targets: True batch_size: 2 shuffle: True drop_last: True ``` 更改后的半监督TrainReader: ```python ### reader config SemiTrainReader: sample_transforms: - Decode: {} - RandomResize: {target_size: [[640, 1333], [672, 1333], [704, 1333], [736, 1333], [768, 1333], [800, 1333]], keep_ratio: True, interp: 1} - RandomFlip: {} weak_aug: - NormalizeImage: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: true} strong_aug: - StrongAugImage: {transforms: [ RandomColorJitter: {prob: 0.8, brightness: 0.4, contrast: 0.4, saturation: 0.4, hue: 0.1}, RandomErasingCrop: {}, RandomGaussianBlur: {prob: 0.5, sigma: [0.1, 2.0]}, RandomGrayscale: {prob: 0.2}, ]} - NormalizeImage: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: true} sup_batch_transforms: - Permute: {} - PadBatch: {pad_to_stride: 32} - Gt2FCOSTarget: object_sizes_boundary: [64, 128, 256, 512] center_sampling_radius: 1.5 downsample_ratios: [8, 16, 32, 64, 128] num_shift: 0.5 multiply_strides_reg_targets: False norm_reg_targets: True unsup_batch_transforms: - Permute: {} - PadBatch: {pad_to_stride: 32} sup_batch_size: 2 unsup_batch_size: 2 shuffle: True drop_last: True ``` ### 其他配置 训练epoch数需要和全量数据训练时换算总iter数保持一致,如全量训练24 epoch(换算约为180k个iter),则10%监督数据的半监督训练,总epoch数需要为240 epoch左右(换算约为180k个iter)。示例如下: ```python ### other config epoch: 240 LearningRate: base_lr: 0.01 schedulers: - !PiecewiseDecay gamma: 0.1 milestones: 240 use_warmup: True - !LinearWarmup start_factor: 0.001 steps: 1000 OptimizerBuilder: optimizer: momentum: 0.9 type: Momentum regularizer: factor: 0.0001 type: L2 clip_grad_by_value: 1.0 ``` ## 使用说明 仅训练时必须使用半监督检测的配置文件去训练,评估、预测、部署也可以按基础检测器的配置文件去执行。 ### 训练 ```bash # 单卡训练 (不推荐,需按线性比例相应地调整学习率) CUDA_VISIBLE_DEVICES=0 python tools/train.py -c ssod/denseteacher/denseteacher_fcos_r50_fpn_coco_semi010.yml --eval # 多卡训练 python -m paddle.distributed.launch --log_dir=denseteacher_fcos_semi010/ --gpus 0,1,2,3,4,5,6,7 tools/train.py -c ssod/denseteacher/denseteacher_fcos_r50_fpn_coco_semi010.yml --eval ``` ### 评估 ```bash CUDA_VISIBLE_DEVICES=0 python tools/eval.py -c ssod/denseteacher/denseteacher_fcos_r50_fpn_coco_semi010.yml -o weights=output/denseteacher_fcos_r50_fpn_coco_semi010/model_final.pdparams ``` ### 预测 ```bash CUDA_VISIBLE_DEVICES=0 python tools/infer.py -c ssod/denseteacher/denseteacher_fcos_r50_fpn_coco_semi010.yml -o weights=output/denseteacher_fcos_r50_fpn_coco_semi010/model_final.pdparams --infer_img=demo/000000014439.jpg ``` ### 部署 部署可以使用半监督检测配置文件,也可以使用基础检测器的配置文件去部署和使用。 ```bash # 导出模型 CUDA_VISIBLE_DEVICES=0 python tools/export_model.py -c ssod/denseteacher/denseteacher_fcos_r50_fpn_coco_semi010.yml -o weights=https://paddledet.bj.bcebos.com/models/denseteacher_fcos_r50_fpn_coco_semi010.pdparams # 导出权重预测 CUDA_VISIBLE_DEVICES=0 python deploy/python/infer.py --model_dir=output_inference/denseteacher_fcos_r50_fpn_coco_semi010 --image_file=demo/000000014439_640x640.jpg --device=GPU # 部署测速 CUDA_VISIBLE_DEVICES=0 python deploy/python/infer.py --model_dir=output_inference/denseteacher_fcos_r50_fpn_coco_semi010 --image_file=demo/000000014439_640x640.jpg --device=GPU --run_benchmark=True # --run_mode=trt_fp16 # 导出ONNX paddle2onnx --model_dir output_inference/denseteacher_fcos_r50_fpn_coco_semi010/ --model_filename model.pdmodel --params_filename model.pdiparams --opset_version 12 --save_file denseteacher_fcos_r50_fpn_coco_semi010.onnx ``` ## 引用 ``` @article{denseteacher2022, title={Dense Teacher: Dense Pseudo-Labels for Semi-supervised Object Detection}, author={Hongyu Zhou, Zheng Ge, Songtao Liu, Weixin Mao, Zeming Li, Haiyan Yu, Jian Sun}, journal={arXiv preprint arXiv:2207.02541}, year={2022} } ```