Skip to content

Latest commit

 

History

History
89 lines (72 loc) · 2.49 KB

README.md

File metadata and controls

89 lines (72 loc) · 2.49 KB

Widerface Benchmark Tutorial

Setup

Install latest version of fastface with

pip install fastface -U

Discovery

fastface is packed with varius pretrained models, to see full list run the following

python -c "import fastface as ff;print(ff.list_pretrained_models())"

Output will be look like

['lffd_original', 'lffd_slim']

Start

Lets import required packages

import fastface as ff
import pytorch_lightning as pl
import torch

Build pretrained model. For this tutorial lffd_original is selected but you can also select another model

model = ff.FaceDetector.from_pretrained("lffd_original")
# model: pl.LightningModule

If you don't have pretrained model weights on your PC, fastface will automatically download and put it under $HOME/.cache/fastface/<package_version>/model/

Add widerface average precision(defined in the widerface competition) metric to the model

metric = ff.metric.WiderFaceAP(iou_threshold=0.5)
# metric: pl.metrics.Metric

# add metric to the model
model.add_metric("widerface_ap", metric)

Define widerface dataset. For this tutorial easy partition is selected but medium or hard partitions are also available
Warning! Do not use batch_size > 1, because tensors can not be stacked due to different size of images. Also using fixed image size drops metric performance.

ds = ff.dataset.WiderFaceDataset(
    phase="test",
    partitions=["easy"],
    transforms= ff.transforms.Compose(
        ff.transforms.ConditionalInterpolate(max_size=1500),
    )
)
# ds: torch.utils.data.Dataset

# get dataloader
dl = ds.get_dataloader(batch_size=1, num_workers=1)
# dl: torch.utils.data.DataLoader

If you don't have widerface validation dataset on your PC, fastface will automatically download and put it under $HOME/.cache/fastface/<package_version>/data/widerface/

Define pytorch_lightning.Trainer

trainer = pl.Trainer(
    benchmark=True,
    logger=False,
    checkpoint_callback=False,
    gpus=1 if torch.cuda.is_available() else 0,
    precision=32)

Run test

trainer.test(model, test_dataloaders=dl)

You should get output like this after test is done

--------------------------------------------------------------------------------
DATALOADER:0 TEST RESULTS
{'widerface_ap': 0.8929094818903156}
--------------------------------------------------------------------------------

Checkout test_widerface.py script to see full code