Charless Fowlkes | University of California, Irvine (original) (raw)
Uploads
Papers by Charless Fowlkes
PLOS Genetics, 2016
Self-renewing organs often experience a decline in function in the course of aging. It is unclear... more Self-renewing organs often experience a decline in function in the course of aging. It is unclear whether chronological age or external factors control this decline, or whether it is driven by stem cell self-renewal-for example, because cycling cells exhaust their replicative capacity and become senescent. Here we assay the relationship between stem cell cycling and senescence in the Caenorhabditis elegans reproductive system, defining this senescence as the progressive decline in "reproductive capacity," i.e. in the number of progeny that can be produced until cessation of reproduction. We show that stem cell cycling diminishes remaining reproductive capacity, at least in part through the DNA damage response. Paradoxically, gonads kept under conditions that preclude reproduction keep cycling and producing cells that undergo apoptosis or are laid as unfertilized gametes, thus squandering reproductive capacity. We show that continued activity is in fact beneficial inasmuch as gonads that are active when reproduction is initiated have more sustained early progeny production. Intriguingly, continued cycling is intermittent-gonads switch between active and dormant states-and in all likelihood stochastic. Other organs face tradeoffs whereby stem cell cycling has the beneficial effect of providing freshly-differentiated cells and the detrimental effect of increasing the likelihood of cancer or senescence; stochastic stem cell cycling may allow for a subset of cells to preserve proliferative potential in old age, which may implement a strategy to deal with uncertainty as to the total amount of proliferation to be undergone over an organism's lifespan.
Occlusion poses a significant difficulty for object recognition due to the combinatorial diversit... more Occlusion poses a significant difficulty for object recognition due to the combinatorial diversity of possible occlusion patterns. We take a strongly supervised, nonparametric approach to modeling occlusion by learning deformable models with many local part mixture templates using large quantities of synthetically generated training data. This allows the model to learn the appearance of different occlusion patterns including figure-ground cues such as the shapes of occluding contours as well as the co-occurrence statistics of occlusion between neighboring parts. The underlying part mixture-structure also allows the model to capture coherence of object support masks between neighboring parts and make compelling predictions of figure-ground-occluder segmentations. We test the resulting model on human pose estimation under heavy occlusion and find it produces improved localization accuracy.
PLOS ONE, 2016
Discriminating between black and white spruce (Picea mariana and Picea glauca) is a difficult pal... more Discriminating between black and white spruce (Picea mariana and Picea glauca) is a difficult palynological classification problem that, if solved, would provide valuable data for paleoclimate reconstructions. We developed an open-source visual recognition software (ARLO, Automated Recognition with Layered Optimization) capable of differentiating between these two species at an accuracy on par with human experts. The system applies pattern recognition and machine learning to the analysis of pollen images and discovers general-purpose image features, defined by simple features of lines and grids of pixels taken at different dimensions, size, spacing, and resolution. It adapts to a given problem by searching for the most effective combination of both feature representation and learning strategy. This results in a powerful and flexible framework for image classification. We worked with images acquired using an automated slide scanner. We first applied a hash-based "pollen spotting" model to segment pollen grains from the slide background. We next tested ARLO's ability to reconstruct black to white spruce pollen ratios using artificially constructed slides of known ratios. We then developed a more scalable hash-based method of image analysis that was able to distinguish between the pollen of black and white spruce with an estimated accuracy of 83.61%, comparable to human expert performance. Our results demonstrate the capability of machine learning systems to automate challenging taxonomic classifications in pollen analysis, and our success with simple image representations suggests that our approach is generalizable to many other object recognition problems.
We present a simple, efficient model for learning boundary detection based on a random forest cla... more We present a simple, efficient model for learning boundary detection based on a random forest classifier. Our approach combines (1) efficient clustering of training examples based on simple partitioning of the space of local edge orientations and (2) scale-dependent calibration of individual tree output probabilities prior to multiscale combination. The resulting model outperforms published results on the challenging BSDS500 boundary detection benchmark. Further, on large datasets our model requires substantially less memory for training and speeds up training time by a factor of 10 over the structured forest model.
BMC bioinformatics, 2015
Analysis of single cells in their native environment is a powerful method to address key question... more Analysis of single cells in their native environment is a powerful method to address key questions in developmental systems biology. Confocal microscopy imaging of intact tissues, followed by automatic image segmentation, provides a means to conduct cytometric studies while at the same time preserving crucial information about the spatial organization of the tissue and morphological features of the cells. This technique is rapidly evolving but is still not in widespread use among research groups that do not specialize in technique development, perhaps in part for lack of tools that automate repetitive tasks while allowing experts to make the best use of their time in injecting their domain-specific knowledge. Here we focus on a well-established stem cell model system, the C. elegans gonad, as well as on two other model systems widely used to study cell fate specification and morphogenesis: the pre-implantation mouse embryo and the developing mouse olfactory epithelium. We report a p...
Local boundary detection remains an important problem in vision. Physiology shows that V1 extract... more Local boundary detection remains an important problem in vision. Physiology shows that V1 extracts complex boundaries (Lee et al, 1998 Vision Research 38 2429-2454). Psychophysics shows that human subjects localise boundaries using multiple cues (Rivest and Cavanagh, 1996 Vision Research 36 53-66). Our work in computer vision (Martin et al, 2002 Neural Information Processing Systems in press) shows how to formulate and combine local boundary cues in natural images. To determine the quality of computational models, ...
The significance of figure-ground organisation was first emphasised by Rubin [1921 Synsoplevede F... more The significance of figure-ground organisation was first emphasised by Rubin [1921 Synsoplevede Figurer (Copenhagen: Gyldendalske)] who noted that size and surroundedness affect the likelihood that a surface is seen as figural. Stimuli designed by Kanizsa and Gerbino [1976, in Vision and Artifact Ed. M Henle (New York: Springer) pp 25-32] and Vecera et al (2002 Journal of Experimental Psychology: General 131 194-205) pointed respectively to the role of convexity and lower-region as cues. We expect that ...
2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2015
Nature protocols, 2015
To facilitate fine-scale phenotyping of whole specimens, we describe here a set of tissue fixatio... more To facilitate fine-scale phenotyping of whole specimens, we describe here a set of tissue fixation-embedding, detergent-clearing and staining protocols that can be used to transform excised organs and whole organisms into optically transparent samples within 1-2 weeks without compromising their cellular architecture or endogenous fluorescence. PACT (passive CLARITY technique) and PARS (perfusion-assisted agent release in situ) use tissue-hydrogel hybrids to stabilize tissue biomolecules during selective lipid extraction, resulting in enhanced clearing efficiency and sample integrity. Furthermore, the macromolecule permeability of PACT- and PARS-processed tissue hybrids supports the diffusion of immunolabels throughout intact tissue, whereas RIMS (refractive index matching solution) grants high-resolution imaging at depth by further reducing light scattering in cleared and uncleared samples alike. These methods are adaptable to difficult-to-image tissues, such as bone (PACT-deCAL), a...
Lecture Notes in Computer Science, 2013
2014 IEEE Conference on Computer Vision and Pattern Recognition, 2014
2014 IEEE Conference on Computer Vision and Pattern Recognition, 2014
Occlusion poses a significant difficulty for object recognition due to the combinatorial diversit... more Occlusion poses a significant difficulty for object recognition due to the combinatorial diversity of possible occlusion patterns. We take a strongly supervised, nonparametric approach to modeling occlusion by learning deformable models with many local part mixture templates using large quantities of synthetically generated training data. This allows the model to learn the appearance of different occlusion patterns including figure-ground cues such as the shapes of occluding contours as well as the co-occurrence statistics of occlusion between neighboring parts. The underlying part mixture-structure also allows the model to capture coherence of object support masks between neighboring parts and make compelling predictions of figure-ground-occluder segmentations. We test the resulting model on human pose estimation under heavy occlusion and find it produces improved localization accuracy.
ACS chemical neuroscience, Jan 20, 2012
Mitochondria, synaptic vesicles, and other cytoplasmic constituents have to travel long distance ... more Mitochondria, synaptic vesicles, and other cytoplasmic constituents have to travel long distance along the axons from cell bodies to nerve terminals. Interruption of this axonal transport may contribute to many neurodegenerative diseases including Alzheimer's disease (AD). It has been recently shown that exposure of cultured neurons to β-amyloid (Aβ) resulted in severe impairment of mitochondrial transport. This Letter describes an integrated microfluidic platform that establishes surface patterned and compartmentalized culture of neurons for studying the effect of Aβ on mitochondria trafficking in full length of axons. We have successfully quantified the trafficking of fluorescently labeled mitochondria in distal and proximal axons using image processing. Selective treatment of Aβ in the somal or axonal compartments resulted in considerable decrease in mitochondria movement in a location dependent manner such that mitochondria trafficking slowed down more significantly proximal...
We describe a model for multi-target tracking based on associating collections of candidate detec... more We describe a model for multi-target tracking based on associating collections of candidate detections across frames of a video. In order to model pairwise interactions between different tracks, such as suppression of overlapping tracks and contextual cues about co-occurence of different objects, we augment a standard min-cost flow objective with quadratic terms between detection variables. We learn the parameters of this model using structured prediction and a loss function which approximates the multi-target tracking accuracy. We evaluate two different approaches to finding an optimal set of tracks under model objective based on an LP relaxation and a novel greedy extension to dynamic programming that handles pairwise interactions. We find the greedy algorithm achieves equivalent performance to the LP relaxation while being 2-7x faster than a commercial solver. The resulting model with learned parameters outperforms existing methods across several categories on the KITTI tracking benchmark.
PLOS Genetics, 2016
Self-renewing organs often experience a decline in function in the course of aging. It is unclear... more Self-renewing organs often experience a decline in function in the course of aging. It is unclear whether chronological age or external factors control this decline, or whether it is driven by stem cell self-renewal-for example, because cycling cells exhaust their replicative capacity and become senescent. Here we assay the relationship between stem cell cycling and senescence in the Caenorhabditis elegans reproductive system, defining this senescence as the progressive decline in "reproductive capacity," i.e. in the number of progeny that can be produced until cessation of reproduction. We show that stem cell cycling diminishes remaining reproductive capacity, at least in part through the DNA damage response. Paradoxically, gonads kept under conditions that preclude reproduction keep cycling and producing cells that undergo apoptosis or are laid as unfertilized gametes, thus squandering reproductive capacity. We show that continued activity is in fact beneficial inasmuch as gonads that are active when reproduction is initiated have more sustained early progeny production. Intriguingly, continued cycling is intermittent-gonads switch between active and dormant states-and in all likelihood stochastic. Other organs face tradeoffs whereby stem cell cycling has the beneficial effect of providing freshly-differentiated cells and the detrimental effect of increasing the likelihood of cancer or senescence; stochastic stem cell cycling may allow for a subset of cells to preserve proliferative potential in old age, which may implement a strategy to deal with uncertainty as to the total amount of proliferation to be undergone over an organism's lifespan.
Occlusion poses a significant difficulty for object recognition due to the combinatorial diversit... more Occlusion poses a significant difficulty for object recognition due to the combinatorial diversity of possible occlusion patterns. We take a strongly supervised, nonparametric approach to modeling occlusion by learning deformable models with many local part mixture templates using large quantities of synthetically generated training data. This allows the model to learn the appearance of different occlusion patterns including figure-ground cues such as the shapes of occluding contours as well as the co-occurrence statistics of occlusion between neighboring parts. The underlying part mixture-structure also allows the model to capture coherence of object support masks between neighboring parts and make compelling predictions of figure-ground-occluder segmentations. We test the resulting model on human pose estimation under heavy occlusion and find it produces improved localization accuracy.
PLOS ONE, 2016
Discriminating between black and white spruce (Picea mariana and Picea glauca) is a difficult pal... more Discriminating between black and white spruce (Picea mariana and Picea glauca) is a difficult palynological classification problem that, if solved, would provide valuable data for paleoclimate reconstructions. We developed an open-source visual recognition software (ARLO, Automated Recognition with Layered Optimization) capable of differentiating between these two species at an accuracy on par with human experts. The system applies pattern recognition and machine learning to the analysis of pollen images and discovers general-purpose image features, defined by simple features of lines and grids of pixels taken at different dimensions, size, spacing, and resolution. It adapts to a given problem by searching for the most effective combination of both feature representation and learning strategy. This results in a powerful and flexible framework for image classification. We worked with images acquired using an automated slide scanner. We first applied a hash-based "pollen spotting" model to segment pollen grains from the slide background. We next tested ARLO's ability to reconstruct black to white spruce pollen ratios using artificially constructed slides of known ratios. We then developed a more scalable hash-based method of image analysis that was able to distinguish between the pollen of black and white spruce with an estimated accuracy of 83.61%, comparable to human expert performance. Our results demonstrate the capability of machine learning systems to automate challenging taxonomic classifications in pollen analysis, and our success with simple image representations suggests that our approach is generalizable to many other object recognition problems.
We present a simple, efficient model for learning boundary detection based on a random forest cla... more We present a simple, efficient model for learning boundary detection based on a random forest classifier. Our approach combines (1) efficient clustering of training examples based on simple partitioning of the space of local edge orientations and (2) scale-dependent calibration of individual tree output probabilities prior to multiscale combination. The resulting model outperforms published results on the challenging BSDS500 boundary detection benchmark. Further, on large datasets our model requires substantially less memory for training and speeds up training time by a factor of 10 over the structured forest model.
BMC bioinformatics, 2015
Analysis of single cells in their native environment is a powerful method to address key question... more Analysis of single cells in their native environment is a powerful method to address key questions in developmental systems biology. Confocal microscopy imaging of intact tissues, followed by automatic image segmentation, provides a means to conduct cytometric studies while at the same time preserving crucial information about the spatial organization of the tissue and morphological features of the cells. This technique is rapidly evolving but is still not in widespread use among research groups that do not specialize in technique development, perhaps in part for lack of tools that automate repetitive tasks while allowing experts to make the best use of their time in injecting their domain-specific knowledge. Here we focus on a well-established stem cell model system, the C. elegans gonad, as well as on two other model systems widely used to study cell fate specification and morphogenesis: the pre-implantation mouse embryo and the developing mouse olfactory epithelium. We report a p...
Local boundary detection remains an important problem in vision. Physiology shows that V1 extract... more Local boundary detection remains an important problem in vision. Physiology shows that V1 extracts complex boundaries (Lee et al, 1998 Vision Research 38 2429-2454). Psychophysics shows that human subjects localise boundaries using multiple cues (Rivest and Cavanagh, 1996 Vision Research 36 53-66). Our work in computer vision (Martin et al, 2002 Neural Information Processing Systems in press) shows how to formulate and combine local boundary cues in natural images. To determine the quality of computational models, ...
The significance of figure-ground organisation was first emphasised by Rubin [1921 Synsoplevede F... more The significance of figure-ground organisation was first emphasised by Rubin [1921 Synsoplevede Figurer (Copenhagen: Gyldendalske)] who noted that size and surroundedness affect the likelihood that a surface is seen as figural. Stimuli designed by Kanizsa and Gerbino [1976, in Vision and Artifact Ed. M Henle (New York: Springer) pp 25-32] and Vecera et al (2002 Journal of Experimental Psychology: General 131 194-205) pointed respectively to the role of convexity and lower-region as cues. We expect that ...
2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2015
Nature protocols, 2015
To facilitate fine-scale phenotyping of whole specimens, we describe here a set of tissue fixatio... more To facilitate fine-scale phenotyping of whole specimens, we describe here a set of tissue fixation-embedding, detergent-clearing and staining protocols that can be used to transform excised organs and whole organisms into optically transparent samples within 1-2 weeks without compromising their cellular architecture or endogenous fluorescence. PACT (passive CLARITY technique) and PARS (perfusion-assisted agent release in situ) use tissue-hydrogel hybrids to stabilize tissue biomolecules during selective lipid extraction, resulting in enhanced clearing efficiency and sample integrity. Furthermore, the macromolecule permeability of PACT- and PARS-processed tissue hybrids supports the diffusion of immunolabels throughout intact tissue, whereas RIMS (refractive index matching solution) grants high-resolution imaging at depth by further reducing light scattering in cleared and uncleared samples alike. These methods are adaptable to difficult-to-image tissues, such as bone (PACT-deCAL), a...
Lecture Notes in Computer Science, 2013
2014 IEEE Conference on Computer Vision and Pattern Recognition, 2014
2014 IEEE Conference on Computer Vision and Pattern Recognition, 2014
Occlusion poses a significant difficulty for object recognition due to the combinatorial diversit... more Occlusion poses a significant difficulty for object recognition due to the combinatorial diversity of possible occlusion patterns. We take a strongly supervised, nonparametric approach to modeling occlusion by learning deformable models with many local part mixture templates using large quantities of synthetically generated training data. This allows the model to learn the appearance of different occlusion patterns including figure-ground cues such as the shapes of occluding contours as well as the co-occurrence statistics of occlusion between neighboring parts. The underlying part mixture-structure also allows the model to capture coherence of object support masks between neighboring parts and make compelling predictions of figure-ground-occluder segmentations. We test the resulting model on human pose estimation under heavy occlusion and find it produces improved localization accuracy.
ACS chemical neuroscience, Jan 20, 2012
Mitochondria, synaptic vesicles, and other cytoplasmic constituents have to travel long distance ... more Mitochondria, synaptic vesicles, and other cytoplasmic constituents have to travel long distance along the axons from cell bodies to nerve terminals. Interruption of this axonal transport may contribute to many neurodegenerative diseases including Alzheimer's disease (AD). It has been recently shown that exposure of cultured neurons to β-amyloid (Aβ) resulted in severe impairment of mitochondrial transport. This Letter describes an integrated microfluidic platform that establishes surface patterned and compartmentalized culture of neurons for studying the effect of Aβ on mitochondria trafficking in full length of axons. We have successfully quantified the trafficking of fluorescently labeled mitochondria in distal and proximal axons using image processing. Selective treatment of Aβ in the somal or axonal compartments resulted in considerable decrease in mitochondria movement in a location dependent manner such that mitochondria trafficking slowed down more significantly proximal...
We describe a model for multi-target tracking based on associating collections of candidate detec... more We describe a model for multi-target tracking based on associating collections of candidate detections across frames of a video. In order to model pairwise interactions between different tracks, such as suppression of overlapping tracks and contextual cues about co-occurence of different objects, we augment a standard min-cost flow objective with quadratic terms between detection variables. We learn the parameters of this model using structured prediction and a loss function which approximates the multi-target tracking accuracy. We evaluate two different approaches to finding an optimal set of tracks under model objective based on an LP relaxation and a novel greedy extension to dynamic programming that handles pairwise interactions. We find the greedy algorithm achieves equivalent performance to the LP relaxation while being 2-7x faster than a commercial solver. The resulting model with learned parameters outperforms existing methods across several categories on the KITTI tracking benchmark.