Gigapixel Imaging

China is doing a version of the DARPA Award, with a focus on gigapixel image detection, tracking, and trajectory prediction. This helps establish China as the leader in a key technology for military and mass surveillance applications.

GigaVision program seeks to revolutionize computer vision when it meets gigapixel videography with both wide field-of-view and high-resolution details.

We host 6 GigaVision challenges (GigaDetection, GigaMOT, GigaTrajectory, GigaReconstruction, GigaRendering and GigaCrowd) with lucrative awards for each regular season to cordially enroll participants from all over the world.

Total Prize: $400,000.00

GigaDetection challenge is to conduct multiple object detection task in gigapixel images, i.e., inferring the bounding boxes of objects of interest with given static gigapixel images.

GigaMOT challenge is to conduct multiple object tracking task in gigapixel videos, i.e., associating objects at different spatial positions and temporal frames with given gigapixel video sequences.

GigaTrajectory challenge is to conduct pedestrian trajectory prediction task in gigapixel videos, i.e., predicting the trajectory of the pedestrian for a certain period of time afterwards, given the short-term trajectory.

They’re offering a new gigapixel corpus:

PANDA is the first gigaPixel-level humAN-centric viDeo dAtaset to support large-scale, long-term, and multi-object visual analysis. The videos in PANDA were captured by gigapixel cameras, covering real-world large-scale scenes with both wide field-of-view (km2 area level) and high resolution details (gigapixel-level/frame), with a great amount of professional labels, including bounding boxes, attributes, trajectories, groups, interactions, etc.

(multiple edits, it was hard to create a copy-pasted summary of the page)


One test of any of these contests is whether someone wins immediately. Sometimes the sponsors don’t actually know how well developed the field is.