visurg/PL-Stitch

PL-Stitch

📚 Paper - 🤖 GitHub

This is the official repository for the paper A Stitch in Time: Learning Procedural Workflow via Self-Supervised Plackett-Luce Ranking.

PL-Stitch is an image foundation model that captures visual changes over time, enabling procedural activity understanding. It takes an image as input and produces a feature vector as output, leveraging the novel Plackett-Luce temporal ranking objective to build a comprehensive understanding of both the static semantic information and the procedural context within each frame.

Star ⭐ us if you like it!

If you use our model or code in your research, please cite our paper:

@misc{che2025stitchtimelearningprocedural,
      title={A Stitch in Time: Learning Procedural Workflow via Self-Supervised Plackett-Luce Ranking}, 
      author={Chengan Che and Chao Wang and Xinyue Chen and Sophia Tsoka and Luis C. Garcia-Peraza-Herrera},
      year={2025},
      eprint={2511.17805},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2511.17805}, 
}

Abstract

Procedural activities, ranging from routine cooking to complex surgical operations, are highly structured as a set of actions conducted in a specific temporal order. Despite their success on static images and short clips, current self-supervised learning methods often overlook the procedural nature that underpins such activities. We expose the lack of procedural awareness in current SSL methods with a motivating experiment: models pretrained on forward and time-reversed sequences produce highly similar features, confirming that their representations are blind to the underlying procedural order. To address this shortcoming, we propose PL-Stitch, a self-supervised framework that harnesses the inherent temporal order of video frames as a powerful supervisory signal. Our approach integrates two novel probabilistic objectives based on the Plackett-Luce (PL) model. The primary PL objective trains the model to sort sampled frames chronologically, compelling it to learn the global workflow progression. The secondary objective, a spatio-temporal jigsaw loss, complements the learning by capturing fine-grained, cross-frame object correlations. Our approach consistently achieves superior performance across five surgical and cooking benchmarks. Specifically, PL-Stitch yields significant gains in surgical phase recognition (e.g., +11.4 pp k-NN accuracy on Cholec80) and cooking action segmentation (e.g., +5.7 pp linear probing accuracy on Breakfast), demonstrating its effectiveness for procedural video representation learning.

🚩 PL-Stitch model

You can download the checkpoint at 🤗 PL-Stitch and run the following code to extract features from your video frames.

import torch
from PIL import Image
from build_model import build_model

# Load the pre-trained pl_stitch model
pl_stitch = build_model(pretrained_weights = 'your path to the model')
pl_stitch.eval()

# Load the image and convert it to a PyTorch tensor
img_path = 'path/to/your/image.jpg'
img = Image.open(img_path)
img = img.resize((224, 224))
img_tensor = torch.tensor(np.array(img)).unsqueeze(0).to('cuda')

# Extract features from the image
outputs = pl_stitch(img_tensor)