insightface人脸关键点检测训练全过程记录（68点+双眼珠）

代码地址：

https://github.com/deepinsight/insightface/tree/master/alignment/synthetics

环境安装

pip install albumentations

pip install torch==1.12.1+cu116 torchvision==0.13.1+cu116 torchaudio==0.12.1 --extra-index-url https://download.pytorch.org/whl/cu116

pip install pytorch-lightning

pip install timm

pip install albumentations

pip install insightface

pip install onnxruntime

训练集

1 下载训练数据集：
Face Synthetics dataset从https://github.com/microsoft/FaceSynthetics下载并将其放在某个地方

2 路径修改：

3 数据准备
使用tools/prepare_synthetics.py进行训练数据准备
过程：

4 数据生成
数据处理完成后会在此文件夹内生成生成的70点标注图片+模型

训练模型

预训练模型、细节修改
预训练模型，不用下载，需要修改点的个数68改为70
使用ResNet50d

root为数据生成后的文件夹路径，填写自己的即可

开始训练
python -u trainer_synthetics.py

训练报错解决

1 报错：

C:\anaconda\envs\insightface-master12\lib\site-packages\torchaudio\backend\utils.py:62: UserWarning: No audio backend is available.
  warnings.warn("No audio backend is available.")
Global seed set to 727
Traceback (most recent call last):
  File "E:\project_mian\insightface-master\alignment\synthetics\trainer_synthetics.py", line 139, in <module>
    cli_main()
  File "E:\project_mian\insightface-master\alignment\synthetics\trainer_synthetics.py", line 101, in cli_main
    train_set = FaceDataset(root_dir=args.root, is_train=True)
  File "E:\project_mian\insightface-master\alignment\synthetics\datasets\dataset_synthetics.py", line 92, in __init__
    A.MotionBlur(blur_limit=(5,12), p=0.1),
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\albumentations\augmentations\blur\transforms.py", line 78, in __init__
    raise ValueError(f"Blur limit must be odd when centered=True. Got: {self.blur_limit}")
ValueError: Blur limit must be odd when centered=True. Got: (5, 12)

解决： 12改为13

2 报错：

type object 'Trainer' has no attribute 'add_argparse_args'
降级并安装 
pip install pytorch-lightning==1.9.4
报错：
Traceback (most recent call last):
  File "E:\project_mian\insightface-master\alignment\synthetics\trainer_synthetics.py", line 139, in <module>
    cli_main()
  File "E:\project_mian\insightface-master\alignment\synthetics\trainer_synthetics.py", line 130, in cli_main
    logger=TensorBoardLogger(osp.join(ckpt_path, 'logs')),
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\pytorch_lightning\loggers\tensorboard.py", line 110, in __init__
    super().__init__(
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\lightning_fabric\loggers\tensorboard.py", line 93, in __init__
    raise ModuleNotFoundError(
ModuleNotFoundError: Neither `tensorboard` nor `tensorboardX` is available. Try `pip install`ing either.

解决：

pip install tensorboardX 

3 报错
type object 'Trainer' has no attribute 'add_argparse_args'
解决

pip install pytorch-lightning==1.9.4

4 报错

 __init__() got an unexpected keyword argument 'progress_bar_refresh_rate'

解决：

注释这行progress_bar_refresh_rate

5 报错

AttributeError: 'Chatbot' object has no attribute 'style'

解决：

pip install gradio==3.50.0 降低版本 先卸载

6 报错：

Traceback (most recent call last):
  File "E:\project_mian\insightface-master\alignment\synthetics\trainer_synthetics.py", line 143, in <module>
    cli_main()
  File "E:\project_mian\insightface-master\alignment\synthetics\trainer_synthetics.py", line 140, in cli_main
    trainer.fit(model, train_loader, val_loader)
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\pytorch_lightning\trainer\trainer.py", line 608, in fit
    call._call_and_handle_interrupt(
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\pytorch_lightning\trainer\call.py", line 38, in _call_and_handle_interrupt
    return trainer_fn(*args, **kwargs)
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\pytorch_lightning\trainer\trainer.py", line 650, in _fit_impl
    self._run(model, ckpt_path=self.ckpt_path)
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\pytorch_lightning\trainer\trainer.py", line 1112, in _run
    results = self._run_stage()
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\pytorch_lightning\trainer\trainer.py", line 1191, in _run_stage
    self._run_train()
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\pytorch_lightning\trainer\trainer.py", line 1214, in _run_train
    self.fit_loop.run()
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\pytorch_lightning\loops\loop.py", line 199, in run
    self.advance(*args, **kwargs)
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\pytorch_lightning\loops\fit_loop.py", line 267, in advance
    self._outputs = self.epoch_loop.run(self._data_fetcher)
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\pytorch_lightning\loops\loop.py", line 199, in run
    self.advance(*args, **kwargs)
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\pytorch_lightning\loops\epoch\training_epoch_loop.py", line 187, in advance
    batch = next(data_fetcher)
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\pytorch_lightning\utilities\fetching.py", line 184, in __next__
    return self.fetching_function()
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\pytorch_lightning\utilities\fetching.py", line 265, in fetching_function
    self._fetch_next_batch(self.dataloader_iter)
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\pytorch_lightning\utilities\fetching.py", line 280, in _fetch_next_batch
    batch = next(iterator)
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\pytorch_lightning\trainer\supporters.py", line 569, in __next__
    return self.request_next_batch(self.loader_iters)
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\pytorch_lightning\trainer\supporters.py", line 581, in request_next_batch
    return apply_to_collection(loader_iters, Iterator, next)
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\lightning_utilities\core\apply_func.py", line 64, in apply_to_collection
    return function(data, *args, **kwargs)
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\torch\utils\data\dataloader.py", line 681, in __next__
    data = self._next_data()
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\torch\utils\data\dataloader.py", line 1376, in _next_data
    return self._process_data(data)
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\torch\utils\data\dataloader.py", line 1402, in _process_data
    data.reraise()
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\torch\_utils.py", line 461, in reraise
    raise exception
RuntimeError: Caught RuntimeError in DataLoader worker process 0.
Original Traceback (most recent call last):
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\torch\utils\data\_utils\worker.py", line 302, in _worker_loop
    data = fetcher.fetch(index)
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\torch\utils\data\_utils\fetch.py", line 52, in fetch
    return self.collate_fn(data)
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\torch\utils\data\_utils\collate.py", line 175, in default_collate
    return [default_collate(samples) for samples in transposed]  # Backwards compatibility.
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\torch\utils\data\_utils\collate.py", line 175, in <listcomp>
    return [default_collate(samples) for samples in transposed]  # Backwards compatibility.
  File "C:\anaconda\envs\insightface-master12\lib\site-packages\torch\utils\data\_utils\collate.py", line 140, in default_collate
    out = elem.new(storage).resize_(len(batch), *list(elem.size()))
RuntimeError: Trying to resize storage that is not resizable

解决：数据尺寸不对应，都改为70即可

开始训练

训练生成模型保存为下图所示：

测试

运行test_synthetics.py

在’data/300W/Validation’中加入pred.txt，txt中为每张图片路径
并替换对应模型路径即可开始测试

测试结果：