21 results about "Object detection" patented technology

A method of detecting objects using multiple sets of object probability data relating to different respective expected object sizes relative to the image size comprises the steps of: (a) storing a list of locations in an image where an object has been detected, the list defining a probability data set and a probability value for each detected object; (b) detecting object locations in a set of object probability data as locations at which a local maximum probability value exceeds a threshold probability value and, if that location is substantially similar to a location previously stored in the list, where the local maximum probability value exceeds that relating to the previously stored location; and (c) repeating step (b) successively in respect of each set of probability data.

The invention discloses a multi-target association tracking method based on continuous maximum energy and an apparent model. The method comprises: first of all, on the basis of object detection and initial locus extraction, by use of an online discrimination apparent model and an AdBoost method, realizing preliminary association of an object, and obtaining a scattered discontinuous tracking locus; then performing secondary association on a short reliable tracking fragment obtained through the preliminary association by use a Hungarian method, wherein though a continuous tracking locus can be obtained through the secondary association, the secondary association is not suitable for a long-term object tracking process; and finally, making up for the defects of the secondary association by use of an intelligent detection method based on maximum energy, and obtaining a final smoother continuous tracking locus.

The invention discloses an SAR (Synthetic Aperture Radar) image object detection method based on a Primal Sketch algorithm, mainly solving the problem that the traditional object detection method cannot realize the detection on different types of artificial objects. The implementing process of the SAR image object detection method based on the Primal Sketch algorithm comprises the following steps of: (1) obtaining a line segment aggregation expressing image structure information by applying the Primal Sketch algorithm to original SAR images; (2) defining and calculating the regularity degree and the regularity ratio of all line segments in the line segment aggregation; (3) determining a seed line segment aggregation for region growing; (4) executing the region growing by taking the seed line segments as the reference to obtain a candidate object region aggregation including artificial objects and natural objects; and (5) screening according to the characteristics of the line segments in the candidate object region aggregation to obtain final artificial objects, namely bridges, ports and buildings. Compared with the prior art, the detection method disclosed by the invention has the advantages of strong applicability, capability of realizing the detection on different types of artificial objects, exact detection result and low false alarm rate; and the invention is suitable for the SAR image object detection under the condition of multiple object types and different sizes of objects.

The present application discloses a method and apparatus for annotating point cloud data. A specific implementation of the method includes: collecting data in a given scenario by using a laser radar and a non-laser radar sensor to respectively obtain point cloud data and sensor data; segmenting and tracking the point cloud data to obtain point cloud segmentation and tracking results; recognizing and tracking feature objects in the sensor data to obtain feature object recognition and tracking results; correcting the point cloud segmentation and tracking results by using the feature object recognition and tracking results, to obtain confidence levels of the point cloud recognition and tracking results; and determining a point cloud segmentation and tracking result whose confidence level is greater than a confidence level threshold as a point cloud annotation result. This implementation reduces the amount of manual work required for annotating point cloud data, thereby reducing the annotation costs.

Methods and apparatus are provided for improved abandoned object recognition using pedestrian detection. An abandoned object is detected in one or more images by determining if one or more detected objects in a foreground of the images comprises a potential abandoned object; applying a trained pedestrian detector to the potential abandoned object to determine if the potential abandoned object comprises at least a portion of a pedestrian; and classifying the potential abandoned object as an abandoned object based on whether the potential abandoned object is not at least a portion of a pedestrian. The trained pedestrian detector is trained using positive training samples comprised of at least portions of human bodies in one or more poses and/or negative training samples comprised of at least portions of abandoned objects.

The invention discloses a human and non-human object identification method based on a PIR (Pyroelectric Infrared) detector, belongs to the technical fields of digital signal processing and mode identification, and particularly relates to infrared intrusion application in a security system. The core thought is based on a general PIR detector, a human and non-human object detection model is established in combination with the characteristic of a bodily form difference between a human and a non-human object, a theoretical basis is laid for the analysis of a signal difference between the human and the non-human object, and a PIR signal characteristic extraction method applicable to human and non-human object identification is provided technically. The PIR signal characteristic extraction method comprises the following steps: 1, performing de-noising and normalization preprocessing on a PIR signal; 2, computing the zero-cross rate of the preprocessed PIR signal; 3, performing first-order difference processing on the preprocessed PIR signal, performing AR (Automatic Regression) model analysis, and solving the regression coefficient of a model by using a Marple algorithm; and 4, applying the zero-cross rate obtained in the step 2 and the regression coefficient obtained in the step 3 to human and non-human object identification as characteristic descriptions of the PIR signal. Lastly, as proved by experiments, the method has high reliability in human and non-human object identification.

A three-dimensional object detection device has an image capturing unit, a three-dimensional object detection unit, a host vehicle speed detection unit, a light source detection unit and a controller. The image capturing unit captures images rearward of a vehicle. The three-dimensional object detection unit detects a presence of a three-dimensional object in a detection area, based on the captured images. The host vehicle speed detection unit detects a vehicle traveling speed. The light source detection unit detects a headlight light source of a headlight of another vehicle. The controller compares the traveling speeds of the object and the vehicle upon not detecting the headlight light source, and suppresses detection of the object upon determining one of the object traveling speed being equal to or less than the vehicle traveling speed, and a difference between the object and vehicle traveling speeds being less than a predetermined value.

The embodiment of the invention relates to a target object tracking method and device based on target object detection and a storage medium. The method comprises: acquiring multiple sets of frame image data; determining position information of a target object in each group of frame image data based on an image detection model; And determining tracking information of a target object in the plurality of groups of frame image data according to the position information. A target detection algorithm is adopted to improve the detection precision and reliability, a frame image detection and trackingmode is adopted to avoid the problem that the target object is lost due to too fast motion of the target object, and compared with a continuous tracking mode, the tracking speed is higher, and the result is more accurate.

The invention discloses a steel bar quantity detection method and system based on a deep neural network. The method comprises the steps: generating a pre-training model according to a third-party object detection data set and based on an improved Faster RCNN algorithm; inputting a steel bar data set including labeling information, performing first data enhancement conversion on pictures in the steel bar data set, and performing transfer learning training on the pre-training model to generate a steel bar detection model; acquiring a to-be-detected picture including the cross section of the steel bar, obtaining a first detection frame through the steel bar detection model, performing second data enhancement conversion on the to-be-detected picture to obtain a plurality of second enhanced pictures, and obtaining a second detection frame through the steel bar detection model; matching the first detection frame and the second detection frame according to the coordinate information, and obtaining the detection number of the steel bars based on the confidence degree of the steel bars and the overlapping degree of the detection frames. The steel bar detection and warehousing speed is increased, manpower and material resources are saved, and the detection accuracy is improved.

The embodiment of the invention discloses an image processing method, device, and equipment and a storage medium. The method comprises the steps of obtaining a to-be-detected target image; performingfeature recognition on the target image to obtain at least one feature map; pooling the at least one feature map to obtain a pooling value of each feature map; performing category prediction on the atleast one feature map to obtain a prediction category of the at least one feature map and a weight value corresponding to each prediction category; determining a thermodynamic diagram corresponding to the target image according to the pooling value of each feature map and the weight value corresponding to the prediction category of each feature map; determining whether a target abnormal object exists in the target image according to the thermodynamic diagram; and if yes, obtaining attribute information of the target abnormal object according to the thermodynamic diagram, and outputting the attribute information of the target abnormal object. According to the embodiment of the invention, abnormal object detection can be better carried out on the image, and the accuracy of a detection result is improved.

A system for detecting objects using ultrasonic waves and methods for making and using the same are provided. The object detection system uniquely encodes each of a plurality of ultrasonic waves and transmit each of the uniquely-encoded ultrasonic waves in a respective direction. The object detection system then receives any of the emitted uniquely-encoded ultrasonic waves that are reflected from an object. By decoding the reflected ultrasonic waves, the object detection system distinguishes among the uniquely-encoded ultrasonic waves and detect the existence and location of the object.

The invention provides a pedestrian identification auxiliary system and a method of a vehicle, belonging to the technical field of automobile object identification. The system and the method address how to improve the accuracy of pedestrian identification. The system includes a camera for photographing an external image of a target vehicle, at least one mobile phone signal detector and a controller. The mobile phone signal detector and the camera are connected to the controller. The method comprises the following steps of: photographing an external image of a target vehicle; detecting the existence of a mobile phone object in an effective area by transmitting a fundamental wave and receiving a feedback signal; identifying the image data of the camera and calculating according to the data sent by the mobile phone signal detector; calculating the distance and the orientation of the mobile phone-carrying object sensed by the correspond target vehicle, and superposing the mobile phone-carrying object determined by the mobile phone signal detector and an image captured by the camera, so as to establish the specific position of the mobile phone-carrying object in the image so as to carryout pedestrian identification. The accuracy of pedestrian or cyclist identification is improved.

The invention discloses an exploring tube clamping and righting device of an object detection well. The exploring tube clamping and righting device comprises a shell, four clamping mechanisms, a base, an adjusting mechanism and a righting mechanism, wherein a through cavity is formed in the shell from top to bottom; the four clamping mechanisms are assembled on the bottom end of an inner cavity of the shell every 90 degrees in the circumferential direction; the base is installed on the top end of the shell; the adjusting mechanism is assembled in the inner cavity of the base; and the righting mechanism is assembled on the top end of the base. According to the device, on one hand, the direction of a fixed base can be adjusted according to the actual hardness of exploration geology, so that the installation of equipment is convenient and rapid, and great convenience is provided for workers; on the other hand, the position of an exploring tube can be rapidly fixed, the operation is simple, and time and labor are saved; and additionally, the exploring tube can be rapidly and accurately righted by the device, the exploring tube is parallel to an exploration hole all the time, the position is stable, the operation is easy, the design is reasonable, the use requirements of the existing market are met, and the device is worthy of popularization.

Detection circuits (1) for detecting movements of movable objects (2) such as joysticks are provided with first detectors (100) for detecting first movements of the joysticks in first directions, comprising first detection units (101) for detecting a presence / absence of light spots (3), locations of the light spots (3) depending on said first movements, and with second detectors (200) for detecting second movements of the joysticks in second directions, comprising second detection units (201) for detecting first / second intensities of the light spots (3), intensities of the light spots (3) depending on said second movements. Such detection circuits (1) are less sensitive to misalignment of components during an assembly and simpler to produce and less costly. The second detectors (200) are entirely located within the light spot (3) independently from positions of the joysticks and the first and third detectors are partly located within the light spot (3) dependently on positions of the joysticks. The detection units (101) comprise photo diodes (120) and transistors (121) for digitizing the signals from the photo diodes (120).

A monitoring apparatus according to the present invention includes: a control unit configured to confirm whether a measured value detected from a monitored object satisfies a preset condition, and execute processing for the monitored object set correspondingly to the condition in a case where the measured value satisfies the condition; and a recording processing unit configured to record an execution status of the processing for the monitored object into preset schedule data.