Paul A. Viola

Bio: Paul A. Viola is an academic researcher from Microsoft. The author has contributed to research in topics: Parsing & Boosting (machine learning). The author has an hindex of 52, co-authored 115 publications receiving 59853 citations. Previous affiliations of Paul A. Viola include IBM & Wilmington University.

Network Based Autonomous Robot Motor Control: from Hormones to Learning

Abstract: Network models have started to be used to control physical robots. A major problem is how to control the selection of grossly different high level behaviors and how to express that selection down at the motor level. In this paper we show how to extend the subsumption architecture [Brooks 86] inspired by a model of behavior selection of [Maes 89], and incorporating a model of hormonal control [Kravitz 88]. In robotic terms a hormonal control system enables global control of a distributed system with only a very low bandwidth global communication system. This has important implications concerning the ability to fabricate intelligent silicon controllers for tiny microfabricated micro robots [Flynn, Brooks and Tavrow 89].

Low resolution optical character recognition for camera acquired documents

Abstract: A global optimization framework for optical character recognition (OCR) of low-resolution photographed documents that combines a binarization-type process, segmentation, and recognition into a single process. The framework includes a machine learning approach trained on a large amount of data. A convolutional neural network can be employed to compute a classification function at multiple positions and take grey-level input which eliminates binarization. The framework utilizes preprocessing, layout analysis, character recognition, and word recognition to output high recognition rates. The framework also employs dynamic programming and language models to arrive at the desired output.

Continuous inference for sequence data

Abstract: Dynamic inference is leveraged to provide online sequence data labeling. This provides real-time alternatives to current methods of inference for sequence data. Instances estimate an amount of uncertainty in a prediction of labels of sequence data and then dynamically predict a label when an uncertainty in the prediction is deemed acceptable. The techniques utilized to determine when the label can be generated are tunable and can be personalized for a given user and/or a system. Employed decoding techniques can be dynamically adjusted to tradeoff system resources for accuracy. This allows for fine tuning of a system based on available system resources. Instances also allow for online inference because the inference does not require knowledge of a complete set of sequence data.

System and method for detecting object in image

Abstract: PROBLEM TO BE SOLVED: To provide a method for detecting an object, such as a human face, in an image. SOLUTION: An image is first partitioned into varisized patches by use of either an integral image or a Gaussian pyramid. Features in each patch are evaluated to determine a cumulative score. The evaluation is repeated as long as the cumulative score is within the range between an acceptance threshold and a rejection threshold, or otherwise the image is rejected when the cumulative score is smaller than the rejection threshold while the image is accepted as the one including the object when the cumulative score is greater than the acceptance threshold. COPYRIGHT: (C)2004,JPO&NCIPI

Learning A* underestimates: Using inference to guide inference

Abstract: We present a technique for speeding up inference of structured variables using a prioritydriven search algorithm rather than the more conventional dynamic programing. A priority-driven search algorithm is guaranteed to return the optimal answer if the priority function is an underestimate of the true cost function. We introduce the notion of a probable approximate underestimate, and show that it can be used to compute a probable approximate solution to the inference problem when used as a priority function. We show that we can learn probable approximate underestimate functions which have the functional form of simpler, easy to decode models. These models can be learned from unlabeled data by solving a linear/quadratic optimization problem. As a result, we get a priority function that can be computed quickly, and results in solutions that are (provably) almost optimal most of the time. Using these ideas, discriminative classifiers such as semi-Markov CRFs and discriminative parsers can be sped up using a generalization of the A* algorithm. Further, this technique resolves one of the biggest obstacles to the use of A* as a general decoding procedure, namely that of coming up with a admissible priority function. Applying this technique results in a algorithm that is more than 3 times as fast as the Viterbi algorithm for decoding semi-Markov Conditional Markov Models.

Histograms of oriented gradients for human detection

Abstract: We study the question of feature sets for robust visual object recognition; adopting linear SVM based human detection as a test case. After reviewing existing edge and gradient based descriptors, we show experimentally that grids of histograms of oriented gradient (HOG) descriptors significantly outperform existing feature sets for human detection. We study the influence of each stage of the computation on performance, concluding that fine-scale gradients, fine orientation binning, relatively coarse spatial binning, and high-quality local contrast normalization in overlapping descriptor blocks are all important for good results. The new approach gives near-perfect separation on the original MIT pedestrian database, so we introduce a more challenging dataset containing over 1800 annotated human images with a large range of pose variations and backgrounds.

You Only Look Once: Unified, Real-Time Object Detection

Abstract: We present YOLO, a new approach to object detection. Prior work on object detection repurposes classifiers to perform detection. Instead, we frame object detection as a regression problem to spatially separated bounding boxes and associated class probabilities. A single neural network predicts bounding boxes and class probabilities directly from full images in one evaluation. Since the whole detection pipeline is a single network, it can be optimized end-to-end directly on detection performance. Our unified architecture is extremely fast. Our base YOLO model processes images in real-time at 45 frames per second. A smaller version of the network, Fast YOLO, processes an astounding 155 frames per second while still achieving double the mAP of other real-time detectors. Compared to state-of-the-art detection systems, YOLO makes more localization errors but is less likely to predict false positives on background. Finally, YOLO learns very general representations of objects. It outperforms other detection methods, including DPM and R-CNN, when generalizing from natural images to other domains like artwork.

Rapid object detection using a boosted cascade of simple features

Abstract: This paper describes a machine learning approach for visual object detection which is capable of processing images extremely rapidly and achieving high detection rates. This work is distinguished by three key contributions. The first is the introduction of a new image representation called the "integral image" which allows the features used by our detector to be computed very quickly. The second is a learning algorithm, based on AdaBoost, which selects a small number of critical visual features from a larger set and yields extremely efficient classifiers. The third contribution is a method for combining increasingly more complex classifiers in a "cascade" which allows background regions of the image to be quickly discarded while spending more computation on promising object-like regions. The cascade can be viewed as an object specific focus-of-attention mechanism which unlike previous approaches provides statistical guarantees that discarded regions are unlikely to contain the object of interest. In the domain of face detection the system yields detection rates comparable to the best previous systems. Used in real-time applications, the detector runs at 15 frames per second without resorting to image differencing or skin color detection.

The Pascal Visual Object Classes (VOC) Challenge

Abstract: The Pascal Visual Object Classes (VOC) challenge is a benchmark in visual object category recognition and detection, providing the vision and machine learning communities with a standard dataset of images and annotation, and standard evaluation procedures. Organised annually from 2005 to present, the challenge and its associated dataset has become accepted as the benchmark for object detection. This paper describes the dataset and evaluation procedure. We review the state-of-the-art in evaluated methods for both classification and detection, analyse whether the methods are statistically different, what they are learning from the images (e.g. the object or its context), and what the methods find easy or confuse. The paper concludes with lessons learnt in the three year history of the challenge, and proposes directions for future improvement and extension.

Fast R-CNN

Abstract: This paper proposes a Fast Region-based Convolutional Network method (Fast R-CNN) for object detection. Fast R-CNN builds on previous work to efficiently classify object proposals using deep convolutional networks. Compared to previous work, Fast R-CNN employs several innovations to improve training and testing speed while also increasing detection accuracy. Fast R-CNN trains the very deep VGG16 network 9x faster than R-CNN, is 213x faster at test-time, and achieves a higher mAP on PASCAL VOC 2012. Compared to SPPnet, Fast R-CNN trains VGG16 3x faster, tests 10x faster, and is more accurate. Fast R-CNN is implemented in Python and C++ (using Caffe) and is available under the open-source MIT License at https://github.com/rbgirshick/fast-rcnn.