Human Activity Detection in Multi-Camera, Continuous, Long-Duration Video (HADCV'19) Workshop

---

---

09:00: Welcome and Introductions

• 09:10: Invited Talk: Detecting Activities with Less, Cees Snoek (University of Amsterdam)

Detecting Activities with Less, Cees Snoek (University of Amsterdam) Abstract

Spatio-temporal detection of human activities in video is demanding in terms of labels and computation. In this talk, I will present recent work that attacks these problems. First, I will discuss supervision for activity detection from video-level class labels only. The state-of-the-art casts this weakly-supervised detection regime as a Multiple Instance Learning problem, where instances are a priori computed spatio-temporal proposals. Rather than disconnecting the spatio-temporal learning from the training, we propose Spatio-Temporal Instance Learning, which enables activity detection directly from box proposals in video frames. Then we will zoom in on representation learning for fully-supervised activity detection. The two-stream detection network based on RGB and flow provides state-of-the-art accuracy at the expense of a large model-size and heavy computation. We propose to embed RGB and optical-flow into a single two-in-one stream network with new layers. A motion condition layer extracts motion information from flow images, which is leveraged by the motion modulation layer to generate transformation parameters for modulating the low-level RGB features. The method is easily embedded in existing appearance- or two-stream action detection networks, and trained end-to-end. Experiments on several video datasets demonstrate the ability to detect human activities with less labels and computation, while maintaining competitive accuracy.

• 09:40: Invited Talk: Activity Detection in Extended Videos - The IARPA DIVA Program, Jeff Alstott (IARPA)

Abstract

Coming Soon .


10:00: Morning Break

ActEV Challenge Results & Best Performer Presentations (1030-1210)
• 10:30: ActEV18: Human Activity Detection Evaluation for Extended Videos, Yooyoung Lee, Jon Fiscus, Afzal Godil, David Joy, Andrew Delgado, Jim Golden
• 10:55: Recent Results from a Proposal-based Approach for Action Detection in Untrimmed Videos, Carlos Castillo
• 11:20: Object-Centric Spatio-Temporal Activity Detection, Quanfu Fan
• 11:45: Novel Activities Detection Algorithm in Extended Videos, Li Yao, Ying Qian

Video Analytics (1210-1300)
• 12:10: Supporting Real-time Public Safety Analytics in the NIST PSCR Public Safety Analytics Portfolio, John Garofolo
• 12:35: A Continuous, Full-scope, Spatio-temporal Tracking Metric based on KL-divergence, Terry Adams
  
13:00: Lunch & Please put up your Posters

Activity Detection/Recognition Session (1430-1700)
• 14:30: Invited Talk: View Invariant and Few Shot Action Recognition, Mubarak Shah (University of Central Florida)


• 15:10: Afternoon Break & Poster Session
• Joint Event Detection and Description in Continuous Video Streams, Huijuan Xu, Boyang Li, Vasili Ramanishka, Leonid Sigal, Kate Saenko
• A Scalable System Architecture for Activity Detection with Simple Heuristics, Rico Thomanek, Christian Roschke, Tony Rolletschke, Manuel Heinzig, Maximilian Eibl, Benny Platte, Robert Manthey, Matthias Vodel, Frank Zimmer, Marc Ritter
• Synthesizing Attributes with Unreal Engine for Fine-grained Activity Analysis, Tae Soo Kim, Mike Peven, Weichao Qiu, Alan Yuille, Gregory D. Hager
• Fine-grained Action Detection in Long Surveillance Videos, Sathyanarayanan Aakur, Daniel Sawyer, Sudeep Sarkar


• 16:15: Minding the Gaps in a Video Action Analysis Pipeline, Jia Chen, Jiang Liu, Junwei Liang, Ting-Yao Hu, Wei Ke, Wayner Barrios, Dong Huang, Alexander Hauptman
• 16:40: ActEV & ActEV-PC Group Discussion

17:00: Closing Remarks

---

In this full day workshop, we will bring all the stakeholders (object detection, tracking, activity detection, pose estimation, machine learning, etc.) together to help advance the state-of-the-art in human activity detection technology in multi-camera video streaming environments. For this workshop, we will present the research finding from the Activities in Extended Video (ActEV) evaluation. We will have invited talks from experts in the field. We will have four talks from the best performers at the ActEV evaluation, and the rest of the performers will be invited to present their work as a poster. In addition, we are inviting the research community to submit unpublished research papers. All accepted papers will be published in the IEEE Digital Library

Activity detection in untrimmed video

Activity detection, recognition, classification and prediction

Multi-camera object detection and tracking in crowded scenes

Human activity and behavior analysis in indoor/outdoor environments for public safety, security, traffic monitoring and control, health, space management, etc.

Human activity monitoring in public spaces

Person re-identification

Human pose estimation and gesture recognition

Human activity understanding

Spatio-temporal activity/object localization

Human behavior and activity analysis

Anomaly detection in indoor/outdoor activities

Human-human and Human-object interaction

Indexing and retrieval of human activity in video datasets

Multi-camera analysis

Evaluation criteria and metrics for Activity detection

Benchmarking datasets and annotations

Machine learning/deep learning methods for activity detection

Big video datasets

Ontology of human Activity

Applications of activity detection and understanding in public safety and security, traffic monitoring and control, crime prevention, human-robot interaction, etc.

As organizers of the workshop we are looking forward to your contributions.
A. Godil, J. Fiscus, T. Adams, A. Hoogs, R. Meth

---

Afzal Godil, NIST
Jonathan G. Fiscus, NIST
Terry Adams, IARPA
Anthony Hoogs, Kitware
Reuven Meth, SAIC
Cees Snoek, University of Amsterdam
Mubarak Shah, University of Central Florida
Jeff Alstott, Program Manager, IARPA Shin'ichi Satoh, National Institute of Informatics
Gregory Hager, JHU
Ajay Divakaran, SRI
Rama Chellappa, UMD
Yi Yao, SRI
Rogerio Feris, IBM
Carlos Castillo, UMD
Yogesh Singh Rawat, UCF
Haider Ali, JHU
Marc Ritter, HS-Mittweida
Michael Ryoo, Indiana University
Mubarak Shah, UCF
Sudeep Sarkar, USF
Fatih Porikli, Australian National University
Snehasis Mukherjee, IIIT
Bo Li, USM
Cees Snoek, University of Amsterdam
Konstantinos Avgerinaki, ITI-Certh
Asim Wagan, JU Jeff Byrne, STR
Yooyoung Lee, NIST
Roddy Collins, Kitware
George Awad, NIST
Asad Butt, NIST

---

For WACV'19 Paper preparation and Author kit, See WACV'19 Submissions page for more details Paper preparation and Author kit.

Workshop Registration
You can register for HADCV'19 workshop at the WACV'19 registration page or by clicking this link: WACV2019 Registration