TY - GEN
T1 - ACCURATE GIGAPIXEL CROWD COUNTING BY ITERATIVE ZOOMING AND REFINEMENT
AU - Bakhtiarnia, Arian
AU - Zhang, Qi
AU - Iosifidis, Alexandros
N1 - Publisher Copyright:
© 2024 IEEE.
PY - 2024
Y1 - 2024
N2 - The increasing prevalence of gigapixel resolutions has presented new challenges for crowd counting. Such resolutions are far beyond the memory and computation limits of current GPUs, and available deep neural network architectures and training procedures are not designed for such massive inputs. Although several methods have been proposed to address these challenges, they are either limited to downsampling the input image to a small size, or borrowing from other gigapixel tasks, which are not tailored for crowd counting. In this paper, we propose a novel method called GigaZoom, which iteratively zooms into the densest areas of the image and refines coarser density maps with finer details. We show that GigaZoom obtains the state-of-the-art for gigapixel crowd counting and improves the accuracy of the next best method by 42%.
AB - The increasing prevalence of gigapixel resolutions has presented new challenges for crowd counting. Such resolutions are far beyond the memory and computation limits of current GPUs, and available deep neural network architectures and training procedures are not designed for such massive inputs. Although several methods have been proposed to address these challenges, they are either limited to downsampling the input image to a small size, or borrowing from other gigapixel tasks, which are not tailored for crowd counting. In this paper, we propose a novel method called GigaZoom, which iteratively zooms into the densest areas of the image and refines coarser density maps with finer details. We show that GigaZoom obtains the state-of-the-art for gigapixel crowd counting and improves the accuracy of the next best method by 42%.
KW - Computer Vision
KW - Crowd Counting
KW - Gigapixel Deep Learning
KW - High-Resolution Deep Learning
UR - http://www.scopus.com/inward/record.url?scp=85195379904&partnerID=8YFLogxK
U2 - 10.1109/ICASSP48485.2024.10446065
DO - 10.1109/ICASSP48485.2024.10446065
M3 - Article in proceedings
AN - SCOPUS:85195379904
SN - 979-8-3503-4486-8
T3 - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
SP - 6510
EP - 6514
BT - ICASSP 2024 - 2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 49th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2024
Y2 - 14 April 2024 through 19 April 2024
ER -