Sven Kreiss, 2019

OpenPifPaf Prediction Demo¶

Programmatically use OpenPifPaf to run multi-person pose estimation on an image.

In [1]:

# Uncomment when using on Google Colab:
# !pip install --upgrade openpifpaf==0.11.1

In [2]:

%matplotlib inline

import io
import numpy as np
import openpifpaf
import PIL
import requests
import torch

device = torch.device('cpu')
# device = torch.device('cuda')  # if cuda is available

print(openpifpaf.__version__)
print(torch.__version__)

0.12b2+5.g5d9a208.dirty
1.7.1

Load an Example Image and Display It¶

Image credit: "Learning to surf" by fotologic which is licensed under CC-BY-2.0.

In [3]:

image_response = requests.get('https://raw.githubusercontent.com/vita-epfl/openpifpaf/main/docs/coco/000000081988.jpg')
pil_im = PIL.Image.open(io.BytesIO(image_response.content)).convert('RGB')
im = np.asarray(pil_im)

with openpifpaf.show.image_canvas(im) as ax:
  pass

Load a Trained Neural Network¶

In [4]:

net_cpu, _ = openpifpaf.network.Factory(checkpoint='shufflenetv2k16').factory()
net = net_cpu.to(device)

openpifpaf.decoder.utils.CifSeeds.threshold = 0.5
openpifpaf.decoder.utils.nms.Keypoints.keypoint_threshold = 0.2
openpifpaf.decoder.utils.nms.Keypoints.instance_threshold = 0.2
processor = openpifpaf.decoder.factory([hn.meta for hn in net.head_nets])

Dataset, Dataloader, Visualizer, Prediction¶

In [6]:

data = openpifpaf.datasets.PilImageList([pil_im])
loader = torch.utils.data.DataLoader(
  data, batch_size=1, pin_memory=True, 
  collate_fn=openpifpaf.datasets.collate_images_anns_meta)

openpifpaf.show.Canvas.show = True
annotation_painter = openpifpaf.show.AnnotationPainter()

for images_batch, _, __ in loader:
  predictions = processor.batch(net, images_batch, device=device)[0]
  with openpifpaf.show.image_canvas(im) as ax:
    annotation_painter.annotations(ax, predictions)

In [ ]: