System and method for on-road traffic density analytics using video stream mining and statistical techniques
AssigneeInfosys Limited
InventorsRudra Narayan Hota, Kishore Jonna, Radha Krishna Pisipati
Life of this patent
Estimating the term…
Abstract
A method and system for analyzing on-road traffic density are provided. The method involves allowing a user to select a video image capturing device and coordinates in a video image frame captured by the video image capturing device such that the coordinates form a region of interest (ROI). The ROI is processed to generate a confidence value and a traffic density value. The traffic density value is compared with a first set of threshold values. Based on the comparison, the traffic density values at different instants in a time window are displayed to enable monitoring of the traffic trend.
Claims50
method for analyzing on-road traffic density comprising:
- receiving, by a traffic management computing device, a user selection of a video image capturing device from among a plurality of video image capturing devices;
- receiving, by the traffic management computing device, a user selection of coordinates in one of one or more video image frames of an on-road traffic scenario captured by the selected video image capturing device such that the coordinates form a closed region of interest;
- the
- segmenting, by the traffic management computing device, the region of interest into one or more overlapping sub-windows;
- converting, by the traffic management computing device, the one or more overlapping sub-windows into one or more feature vectors through a textural feature extraction technique;
- generating, by the traffic management computing device, at least a traffic confidence value or no traffic confidence value for each of the feature vectors to classify the sub-windows as having a high traffic value or a low traffic value by a traffic density classifier;
- computing, by the traffic management computing device, at least a traffic density value depending on a number of the sub-windows with a high traffic value and a total number the of sub-windows within the region of interest;
- comparing, by the traffic management computing device, the traffic density value with a first set of threshold values to categorize the video image frame as having low, medium or high traffic;
- and
- displaying, by the traffic management computing device, traffic density values at different instants in a time window to enable monitoring of a traffic trend.
A method for re-training a traffic density classifier comprising:
- collecting, by a traffic management computing device, a set of misclassified video image frames captured by an image capturing device from among a plurality of image capturing devices;
- and
- utilizing, by the traffic management computing device;
- a reinforcement learning to train the traffic density classifier with a valid set of video image frames corresponding to predefined settings of the image capturing device.
road traffic management computing device comprising:
- a processor coupled to a memory and configured to execute programmed instructions stored in the memory, comprising:
- receiving a user selection of a video image capturing device from among a plurality of video image capturing devices communicatively coupled to the traffic management computing device:
- receiving a user selection of coordinated in one of one or more video image frames of an on-road traffic scenario captured by the selected video image capturing device such that the coordinates form a closed region of interest;
- segmenting the region of interest into on or more overlapping sub-windows;
- converting the one or more overlapping sub-windows into one or more feature vectors through a textural feature extraction technique;
- generating at least a traffic confidence value or no traffic confidence value for each of the feature vectors to classify the sub-windows as having at least a high traffic value or a low traffic value by a traffic density classifier;
- computing a traffic density value depending on a number of the sub-windows with a high traffic value and a total number the sub-windows within the region of interest;
- comparing the traffic density value with a first set of threshold values to categorize the video image frame as having low, medium or high traffic;
- and
- displaying traffic density values at different instants in a time window to enable monitoring of a traffic end.
A traffic management computing device comprising:
- a processor coupled to a memory and configured to execute programmed instructions stored in the memory, comprising:
- collecting a set of misclassified video image data of a video image capturing device from among plurality of video image capturing devices;
- and
- utilizing a reinforcement learning to train a traffic density classifier with a valid set of video image data for corresponding to predefined settings of the video image capturing devices.
A non-transitory computer readable medium program having stored thereon instructions for analyzing on-road traffic density comprising machine executable code which when executed by a processor, causes the processor to perform steps comprising:
- receiving a user selection of a video image capturing device from among a plurality of video image capturing devices communicatively coupled to the traffic management computing device;
- receiving a user selection of coordinated in one of one or more video image frames of an on-road traffic scenario captured by the selected video image capturing device such that the coordinates form a closed region of interest;
- segmenting the region of interest into on or more overlapping sub-windows;
- converting the one or more overlapping sub-windows into one or more feature vectors through a textural feature extraction technique;
- generating at least a traffic confidence value or no traffic confidence value for each of the feature vectors to classify the sub-windows as having at least a high traffic value or a low traffic value by a traffic density classifier;
- computing a traffic density value depending on a number of the sub-windows with a high traffic value and a total number the sub-windows within the region of interest;
- comparing the traffic density value with a first set of threshold values to categorize the video image frame as having low, medium or high traffic;
- and
- displaying traffic density values at different instants in a time window to enable monitoring of a traffic end.
A non-transitory computer readable medium program having stored thereon instructions for re-training a traffic density classifier comprising machine executable code which when executed by a processor, causes the processor to perform steps comprising:
- collecting a set of misclassified video image frames captured by an image capturing device from among a plurality of image capturing devices;
- and
- utilizing a reinforcement learning to train the traffic density classifier with a valid set of video image frames corresponding to predefined settings of the image capturing device.
Dependencies are inferred from the claim text.