discussion / AI for Conservation  / 5 June 2019

Workshop & Challenges on Computer Vision for Wildlife Conservation (CVWC) at ICCV 2019

Please consider submitting to this workshop and/or challenge which may be relevant to your work:

 

Call for Papers and Call for challenge participants: Workshop & Challenges on Computer Vision for wildlife conservation (CVWC) at ICCV 2019

 

###Introduction

Conservation of wildlife is important to maintaining a healthy and balanced ecosystem, and ensuring the continued biodiversity of our world. In particular, endangered species serve as an important indicator for biodiversity and environmental health. Governments around the world along with environmental organizations such as WWF (World Wildlife Fund) have dedicated many resources and projects to protect endangered species.

 

Fortunately, computer vision techniques have shown promises in addressing several of these challenges, since we are now able to collecting plenty of imagery data from camera trap or even UAVs, and use this imagery to build edge-to-cloud systems for wildlife conservation. From the edge deployment perspective, CV techniques can be applied to smart imaging sensors to capture wildlife related images/video and monitor wildlife. Because cloud systems have access to significantly more compute, we can apply more sophisticated tasks such as re-identifying certain wildlife individuals from a large amount of photos from distributed cameras, tracking movement patterns, and aggregating population information across multiple sensors.

 

This workshop aims to enhance the social responsibility of the CV community, and bring together researchers in the community to advance wildlife conservation using CV techniques from 3 aspects:

-Welcome contributed papers in a broad area of CV for wildlife conservation.

-Organize a challenge on dataset we collected for Amur tiger conservation with tasks like tiger detection, pose estimation and re-identification.

-Foster new ideas and directions on "CV for wildlife conservation" with invited talks and panel discussions from both the CV community and traditional wildlife conservation community.

 

###Call for papers

Workshop topics include (but are not limited to):

-Fine-grained wildlife recognition

-Wildlife re-identification

-Wildlife tracking

-Smart trap sensor design

-Drone based monitoring/tracking

-Simulation and visualization

-Full conservation system

-Dataset related to wildlife conservation

-Challenge solutions

 

We invite paper submissions and featured works in three tracks:

-Contributed original papers, acceptance based on peered review. 

-Featured existing peer-reviewed papers in the passed 1~2 years

-Challenge solution papers

 

###Challenge

Based on Amur Tiger Re-identification in Wild (ATRW) dataset, we will host a challenge containing following tracks:

 

-Track1: Tiger bounding box Detection

-Track2: Tiger Pose Keypoint Detection

-Track3: Tiger Re-ID: train and test with manually annotated bounding boxes/keypoints

-Track4:Tiger Re-ID in the Wild: re-ID based on automatical detector/keypoint-pose outputs.

The workshop will provide awards for each challenge track winner team thanks to our sponsor's generous donation.

 

###Organizers

-Jianguo Li, Intel Labs

-Weiyao Lin, Shanghai Jiao Tong University

-Hanlin Tang, Intel AI Lab

-Greg Mori, Simon Fraser University

-Joachim Denzler, Friedrich Schiller University Jena

 

###Advisory Board

-Yoshua Bengio, Professor, MILA

-Pietro Perona, Professor, Caltech

-Lucas Joppa, Chief Environmental Officer, Microsoft

-Zhengyou Zhang, Director, Tencent AI Lab

 

###Important Dates

-Training/validation dataset + development kit released June 28, 2019

-Testing dataset released July 26, 2019

-Contributed Paper submission July 31, 2019

-Result submission August 2, 2019

-Challenge result notification August 9, 2019

-Challenge Paper submission August 15, 2019

-Acceptance notification August 23, 2019

-Camera-ready August 30, 2019

-Workshop October 27, 2019

 

### Workshop Website

https://cvwc2019.github.io/