Microsoft Technology Licensing, LLC

Google Inc.

Other

King Abdulaziz City for Science and Technology

Ahmad A Abdulkader

Qi Zhang

Hartmut Neven

Hartwig Adam

Luc Vincent

Andrea Lynn Frome

Bo Wu

Marco Zennaro

Henry Allan Rowley

Gregory Nicholas Hullender

Alessandro Bissacco

Michael T Black

Ioannis A Drakopoulos

German Cheung

Bodin Dresevic

Paul A Viola

John Grant Bennett

Kumar Hemachandra Chellapilla

Deepak Agarwal

Matthew Ryan Casey

Goran Predovic

Sanaz Ahari

Mohammad Sulaiman Khorsheed

Michael David Revow

Girish Kumar

Farid Hosseini

Aparna Lakshmiratan

Angshuman Guha

Kong Man Cheung

Hussein Khalid Al-Omari

Alnur Ali

Ankur Gupta

Milan Vukosavljevic

Brian Leung

Petr Slavik

Vijay Nair

Sameer Yusufali Merchant

Patrice Y Simmard

Iaonnis A Drakopoulos

Nazan Khan

Giridhar Kumaran

Seyed Ali Ahmadi

Alessandro Bissallo

Computer-readable media, computer systems, and computing methods are provided for classifying search results as either of good quality or of poor quality. Initially, a portion of the search results, such as the highest ranked documents, are selected for evaluation. A level of quality for each of the selected search results is determined using a classification process that includes the following steps: targeting features demonstrated by the selected search results to be evaluated; evaluating the selected features to generate a level-of-quality score for each of the selected search results; comparing the score against a predefined threshold value; and, based on the comparison, assigning each of the selected search results an absolute measurement. The absolute measurement indicates poor quality when the score is less than the threshold value. Upon recognizing that the selected search results are of poor quality, automatically executing a corrective action that reformulates the issued search query.

Classifying results of search queries

Embodiments of this invention relate to detecting and blurring images. In an embodiment, a system detects objects in a photographic image. The system includes an object detector module configured to detect regions of the photographic image that include objects of a particular type at least based on the content of the photographic image. The system further includes a false positive detector module configured to determine whether each region detected by the object detector module includes an object of the particular type at least based on information about the context in which the photographic image was taken.

Object detection with false positive filtering

A method and apparatus for automatically identifying character segments for character recognition is provided. The method involves receiving a plurality of words and a ground truth corresponding to each word of the plurality of words. The plurality of words may be received in a cursive script. Each word of the plurality of words is segmented into one or more character segments based on the ground truth corresponding to each word. Thereafter, the segmentation of each word is refined by iteratively re-segmenting each word based on one or more similar character segments.

Method and apparatus for automatically identifying character segments for character recognition

Systems, methods, and computer storage media having computer-executable instructions embodied thereon for rewriting queries and labeling word pairs. Queries are received and alternate words are identified for word pairs (i.e., query words and alternate words). Word pair links are presented to users and indicators are received based on actions taken by the users. Labels are assigned to the word pairs based on the indicators and communicated to a classifier.

Efficient label acquisition for query rewriting

OCR errors are identified and corrected through learning. An error probability estimator is trained using ground truths to learn error probability estimation. Multiple OCR engines process a text image, and convert it into texts. The error probability estimator compares the outcomes of the multiple OCR engines for mismatches, and determines an error probability for each of the mismatches. If the error probability of a mismatch exceeds an error probability threshold, a suspect is generated and grouped together with similar suspects in a cluster. A question for the cluster is generated and rendered to a human operator for answering. The answer from the human operator is then applied to all suspects in the cluster to correct OCR errors in the resulting text. The answer is also used to further train the error probability estimator.

Efficient identification and correction of optical character recognition errors through learning in a multi-engine environment

Various technologies and techniques are disclosed that improve handwriting recognition operations. Handwritten input is received in training mode and run through several base recognizers to generate several alternate lists. The alternate lists are unioned together into a combined alternate list. If the correct result is in the combined list, each correct/incorrect alternate pair is used to generate training patterns. The weights associated with the alternate pairs are stored. At runtime, the combined alternate list is generated just as training time. The trained comparator-net can be used to compare any two alternates in the combined list. A template matching base recognizer is used with one or more neural network base recognizers to improve recognition operations. The system provides comparator-net and reorder-net processes trained on print and cursive data, and ones that have been trained on cursive-only data. The respective comparator-net and reorder-net processes are used accordingly.

Combiner for improving handwriting recognition

In embodiments consistent with the subject matter of this disclosure, a user may input strokes as digital ink to a processing device. The processing device may partition the input strokes into multiple regions of strokes. A first recognizer and a second recognizer may score grammar objects included in regions and represented by chart entries. The scores may be converted to a converted score, which may have at least a near standard normal distribution. The processing device may present a recognition result based on highest converted scores according to a recurrence formula. The processing device may receive a correction hint with respect to misrecognized strokes and may add a penalty score with respect to chart entries representing grammar objects breaking the correction hint. Incremental recognition may be performed when a pause is detected during inputting of strokes.

Recognition of mathematical expressions

Handwriting recognition techniques employing a personalized handwriting recognition engine. The recognition techniques use examples of an individual's previous writing style to help recognize new pen input from that individual. The techniques also employ a shape trainer to select samples of an individual's handwriting that accurately represent the individual's writing style, for use as prototypes to recognize subsequent handwriting from the individual. The techniques also alternately or additionally employ an intelligent combiner to combine the recognition results from the personalized recognition engine and the conventional recognition engine (or engines). The combiner may use a comparative neural network to combine the recognition results from multiple recognition engines. The combiner alternately may use a rule-based system based on prior knowledge of different recognition engines.

Personalized implicit and explicit character shape adaptation and recognition

Systems and methods that exploit unique properties of a language script (e.g., condition joining rules for Arabic language) to enable a two tier text recognition. In such two tier system, one tier can recognize predetermined groups of linked letters that are connected based on joining rules of a language associated with the text, and another tier dissects (and recognizes) such linked letters to respective constituent letters that form the predetermined group of linked letters. Various classifiers and artificial intelligence components can further facilitate text recognition at each level.

Two tiered text recognition

Techniques for processing handwriting input based upon a user's writing style. Some techniques employ the style in which the user writes a single character, while other techniques alternately or additionally employ a group of allographs that form a handwriting style. Some implementations of these techniques, such as those implemented in writing style analysis tool, analyze one or more characters written by a user to identify a community, such as a geographic region or cultural group, to which the user's handwriting style belongs. Other implementations analyze one or more characters of a user's handwriting in order to alternately or additionally categorize the user's handwriting into a particular handwriting style. The writing style analysis tool may then provide the user with a handwriting recognition application specifically configured for that user's personal handwriting style. With still other implementations, the user's handwriting style alternately or additionally is employed to prevent a handwriting recognition application from using an aberrantly written character as training data for improving the recognition process, or to predict how a user writes other characters.

Style aware use of writing input

The claimed subject matter provides a system and/or a method that facilitates analyzing and/or recognizing a handwritten character. An interface component can receive at least one handwritten character. A personalization component can train a classifier based on an allograph related to a handwriting style to provide handwriting recognition for the at least one handwritten character. In addition, the personalization component can employ any suitable combiner to provide optimized recognition.

Allograph based writer adaptation for handwritten character recognition

Systems and methods are provided for training neural networks and other systems with heterogeneous data. Heterogeneous data are partitioned into a number of data categories. A user or system may then assign an importance indication to each category as well as an order value which would affect training times and their distribution (higher order favoring larger categories and longer training times). Using those as input parameters, the ordered training generates a distribution of training iterations (across data categories) and a single training data stream so that the distribution of data samples in the stream is identical to the distribution of training iterations. Finally, the data steam is used to train a recognition system (e.g., an electronic ink recognition system).

Training with heterogeneous data

Growing community of inventors

Woodinville, WA, United States of America

Ahmad A Abdulkader