m monad-knowledge FIIT STU · open research vault
dataset

ucsd

Also known as: UCSD, UCSD dataset, UCSD Pedestrian, UCSD anomaly detection, UCSD Anomaly Detection Dataset, UCSD Ped1, UCSD Ped2

Description

The UCSD dataset (Mahadevan et al., UCSD) is a fixed-camera video benchmark used for both crowd counting and crowd anomaly detection. It is split into two subsets at resolutions of 158x238 and 240x360 pixels (commonly referred to as UCSD Ped1 and Ped2) capturing pedestrian walkways at the UCSD campus. Like Mall, it is a single continuous video sequence per scene, which limits perspective variability.

Modality / size

  • Modality: RGB video from a fixed surveillance camera.
  • Subjects / scenarios: campus pedestrian walkways; two subsets at different resolutions.
  • Labels: anomaly annotations and per-frame pedestrian counts.

Used by (papers)

  • Crowd anomaly detection benchmark; also a low-density crowd-counting baseline alongside Mall.

Notes

  • Public dataset; UCSD Statistical Visual Computing Lab release.
  • Both UCSD Ped1 and UCSD Ped2 are folded into this note since they share the collection setup; the subset distinction lives in the alias list.
Where it appears

4 vault papers evaluate on this dataset

Titles and DOIs only — no abstracts, no analyses.

  • A survey of recent advances in CNN-based single image crowd counting and density estimation 2018 DOI ↗
  • Data-driven Crowd Modeling Techniques: A Survey 2022 DOI ↗
  • Constructing WiFi-Video-Fused Multi-Modal Synthetic Datasets for Crowd Counting 2025 DOI ↗
  • Intelligent video surveillance: a review through deep learning techniques for crowd analysis 2019 DOI ↗
Co-mentioned with
Anomaly Detection×3 WorldExpo '10×3 Crowd Monitoring×3 Crowd Density Estimation×3 Deep Learning×2 Support Vector Machine×2 Camera×2 ShanghaiTech×2 Crowd Counting×2 Pedestrian Dynamics Simulation×1