-
SynCellFactory: Generative Data Augmentation for Cell Tracking
Authors:
Moritz Sturm,
Lorenzo Cerrone,
Fred A. Hamprecht
Abstract:
Cell tracking remains a pivotal yet challenging task in biomedical research. The full potential of deep learning for this purpose is often untapped due to the limited availability of comprehensive and varied training data sets. In this paper, we present SynCellFactory, a generative cell video augmentation. At the heart of SynCellFactory lies the ControlNet architecture, which has been fine-tuned t…
▽ More
Cell tracking remains a pivotal yet challenging task in biomedical research. The full potential of deep learning for this purpose is often untapped due to the limited availability of comprehensive and varied training data sets. In this paper, we present SynCellFactory, a generative cell video augmentation. At the heart of SynCellFactory lies the ControlNet architecture, which has been fine-tuned to synthesize cell imagery with photorealistic accuracy in style and motion patterns. This technique enables the creation of synthetic yet realistic cell videos that mirror the complexity of authentic microscopy time-lapses. Our experiments demonstrate that SynCellFactory boosts the performance of well-established deep learning models for cell tracking, particularly when original training data is sparse.
△ Less
Submitted 26 April, 2024; v1 submitted 25 April, 2024;
originally announced April 2024.
-
CellTypeGraph: A New Geometric Computer Vision Benchmark
Authors:
Lorenzo Cerrone,
Athul Vijayan,
Tejasvinee Mody,
Kay Schneitz,
Fred A. Hamprecht
Abstract:
Classifying all cells in an organ is a relevant and difficult problem from plant developmental biology. We here abstract the problem into a new benchmark for node classification in a geo-referenced graph. Solving it requires learning the spatial layout of the organ including symmetries. To allow the convenient testing of new geometrical learning methods, the benchmark of Arabidopsis thaliana ovule…
▽ More
Classifying all cells in an organ is a relevant and difficult problem from plant developmental biology. We here abstract the problem into a new benchmark for node classification in a geo-referenced graph. Solving it requires learning the spatial layout of the organ including symmetries. To allow the convenient testing of new geometrical learning methods, the benchmark of Arabidopsis thaliana ovules is made available as a PyTorch data loader, along with a large number of precomputed features. Finally, we benchmark eight recent graph neural network architectures, finding that DeeperGCN currently works best on this problem.
△ Less
Submitted 17 May, 2022;
originally announced May 2022.
-
End-to-End Learned Random Walker for Seeded Image Segmentation
Authors:
Lorenzo Cerrone,
Alexander Zeilmann,
Fred A. Hamprecht
Abstract:
We present an end-to-end learned algorithm for seeded segmentation. Our method is based on the Random Walker algorithm, where we predict the edge weights of the underlying graph using a convolutional neural network. This can be interpreted as learning context-dependent diffusivities for a linear diffusion process. Besides calculating the exact gradient for optimizing these diffusivities, we also p…
▽ More
We present an end-to-end learned algorithm for seeded segmentation. Our method is based on the Random Walker algorithm, where we predict the edge weights of the underlying graph using a convolutional neural network. This can be interpreted as learning context-dependent diffusivities for a linear diffusion process. Besides calculating the exact gradient for optimizing these diffusivities, we also propose simplifications that sparsely sample the gradient and still yield competitive results. The proposed method achieves the currently best results on a seeded version of the CREMI neuron segmentation challenge.
△ Less
Submitted 22 May, 2019;
originally announced May 2019.
