NightOwls Pedestrian Detection Challenge 2019

Organized by lukeN86 - Current server time: Dec. 5, 2019, 5:45 p.m. UTC

Previous

Testing Phase
Sept. 30, 2019, midnight UTC

Current

Testing Phase
Sept. 30, 2019, midnight UTC

End

Competition Ends
Oct. 13, 2019, 11:59 p.m. UTC

NightOwls Pedestrian Detection Challenge

Computer Vision for Road Scene Understanding and Autonomous Driving Workshop, ICCV 2019

Pedestrian detection at night from RGB camera is an under-represented yet very important problem, where current state-of-the-art vision algorithms fail. Computer vision methods for detection at night have not received much attention, despite the fact they are a critical building block of many systems such as safe and robust autonomous cars.

To further assess and advance the state of the art, we organize the NightOwls Pedestrian Detetection Challenge 2019, as part of the Computer Vision for Road Scene Understanding and Autonomous Driving (CVRSUAD) Workshop at ICCV 2019.

The competition uses the recently published NightOwls dataset, consisting of 279,000 fully-annotated images in 40 video sequences recorded at night across 3 different countries by an industry-standard camera. Participants are encouraged to train their models on the training subset (128k images), tune the hyper-parameters on the validation subset (48k) and then submit their detection results on the testing subset (128k images). Data annotations are available for the training/validation subset, but the annotations for the testing set will only be published after the competition ends, to ensure fair competition.

The winner will be announced at the CVRSUAD workshop, ICCV 2019 on 27th October and will be presented with valuable and unique prices, by courtesy of Visual Geometry Group, University of Oxford :) After that, the submission site will switch to continous mode.

Dates

  • Sept 2019 - training and validation sets published (images + annotations)
  • 30.09.2019 - testing set published (images only), submission site opens for competition entries
  • 13.10.2019 23:59 UTC - competition entries submission deadline
  • 27.10.2019 - winners announced at CVRSUAD workshop, ICCV 2019

Competition Organizers
nightowls-competition@googlegroups.com

Evaluation

Methods ranking is based on the standard Average Miss Rate metric used in the pedestrian detection literature [1], considering only targets within the Reasonable [1] set up (i.e. non-occluded targets with height >= 50px). The winning entry will be the method with the lowest Average Miss Rate.

The server expects a single ZIP archive with a single JSON file inside. The server runs the same evaluation code from our NightOwls SDK, the evaluation can therefore easily also be run locally.

1. P. Dollár, C. Wojek, B. Schiele and P. Perona, Pedestrian Detection: A Benchmark, CVPR 2009, Miami, Florida

Terms and Conditions

This dataset is made freely available to academic and non-academic entities for non-commercial purposes such as academic research, teaching, scientific publications, or personal experimentation. Permission is granted to use the data given that you agree:

  1. That the dataset comes “AS IS”, without express or implied warranty. Although every effort has been made to ensure accuracy, we (University of Oxford) do not accept any responsibility for errors or omissions.
  2. That you include a reference to the Nightowls Dataset in any work that makes use of the dataset.
  3. That you do not distribute this dataset or modified versions. It is permissible to distribute derivative works in as far as they are abstract representations of this dataset (such as models trained on it or additional annotations that do not directly include any of our data) and do not allow to recover the dataset or something similar in character.
  4. You may not use the dataset or any derivative work for commercial purposes such as, for example, licensing or selling the data, or using the data with a purpose to procure a commercial gain.
  5. That all rights not expressly granted to you are reserved by us (University of Oxford).

Training/Validation Phase

Start: Sept. 1, 2019, midnight

Testing Phase

Start: Sept. 30, 2019, midnight

Competition Ends

Oct. 13, 2019, 11:59 p.m.

You must be logged in to participate in competitions.

Sign In
# Username Score
1 hongsong.wang 0.078861
2 nemonameless 0.113212
3 westbrook 0.113265