Public articles linked to the same research event.
arXiv The authors present FloVMos, an optical-flow-based deep learning video mosaicking framework that fine-tunes an optical flow network on synthetic training data with ground-truth deformation fields, generated automatically from existing large-FOV mosaics or raw videos to adapt to different imaging modalities; across seven modalities (reflection confocal microscopy, open-top light-sheet microscopy, fetoscopy, laparoscopy, dermoscopy, sparse spectral microscopy, and endoscopy), FloVMos outperformed the Parallax, APAP, and AVM baselines in accuracy, robustness, and speed, preserving feature-point distances with an average error of about 0.5% of the field of view and processing 1 MP images at roughly ten frames per second.
The authors present FloVMos, an optical-flow-based deep learning video mosaicking framework that fine-tunes an optical flow network on synthetic training data with ground-truth deformation fields, generated automatically from existing large-FOV mosaics or raw videos to adapt to different imaging modalities; across seven modalities (reflection confocal microscopy, open-top light-sheet microscopy, fetoscopy, laparoscopy, dermoscopy, sparse spectral microscopy, and endoscopy), FloVMos outperformed the Parallax, APAP, and AVM baselines in accuracy, robustness, and speed, preserving feature-point distances with an average error of about 0.5% of the field of view and processing 1 MP images at roughly ten frames per second.
The authors present FloVMos, an optical-flow-based deep learning video mosaicking framework that fine-tunes an optical flow network on synthetic training data with ground-truth deformation fields, generated automatically from existing large-FOV mosaics or raw videos to adapt to different imaging modalities; across seven modalities (reflection confocal microscopy, open-top light-sheet microscopy, fetoscopy, laparoscopy, dermoscopy, sparse spectral microscopy, and endoscopy), FloVMos outperformed the Parallax, APAP, and AVM baselines in accuracy, robustness, and speed, preserving feature-point distances with an average error of about 0.5% of the field of view and processing 1 MP images at roughly ten frames per second.
The authors present FloVMos, an optical-flow-based deep learning video mosaicking framework that fine-tunes an optical flow network on synthetic training data with ground-truth deformation fields, generated automatically from existing large-FOV mosaics or raw videos to adapt to different imaging modalities; across seven modalities (reflection confocal microscopy, open-top light-sheet microscopy, fetoscopy, laparoscopy, dermoscopy, sparse spectral microscopy, and endoscopy), FloVMos outperformed the Parallax, APAP, and AVM baselines in accuracy, robustness, and speed, preserving feature-point distances with an average error of about 0.5% of the field of view and processing 1 MP images at roughly ten frames per second.