7results about How to "Eliminate false detections" patented technology

This invention discloses a pedestrian detection method and system based on head center point-assisted feature comparison. The method includes: constructing a new CSP network architecture by adding a head branch, a feature comparison branch, a detection confidence enhancer, and a head-aware NMS module to the original CSP network architecture; constructing a training sample set, where the head center point is used as the positive sample for the head branch, and non-head center points within the head bounding box are used as the negative sample for the head branch; using the training sample set as input to the CSP network for training to obtain a pedestrian detection network model; supervising network learning during the training phase using head branch loss and feature comparison loss; and enhancing the detection results during the training and testing phases using the detection confidence enhancer and the head-aware NMS method. This invention improves the detection accuracy of CSP and reduces the probability of false positives and false negatives by introducing head center point, feature comparison, detection confidence enhancer, and head-aware NMS methods.

This invention relates to the field of industrial visual intelligent inspection technology, and in particular to an intelligent visual inspection system for surface defects of label paper. The system continuously acquires multiple frames of images of the same label area during roll-to-roll feeding, establishes spatial correspondence between adjacent frames, and calculates the time-series changes in grayscale distribution, edge position, and texture features of the divided structural regions. Through continuous discrimination, progressively changing areas are identified as process evolution areas. Areas that do not meet the continuous change pattern are anomaly screened and defect identification signals are generated. This solution changes the processing method based on single-frame matching tolerance for defect judgment, maintaining anomaly identification stability under tension fluctuations or local deformation conditions, and is suitable for detecting surface defects of various types of labels.

This application relates to the field of vehicle intelligent control technology, and provides an obstacle detection method, a self-moving device, and a storage medium. The method includes: acquiring obstacle detection information collected by each ultrasonic sensor in an ultrasonic sensor group fixedly mounted on the self-moving device; wherein the ultrasonic sensor group includes a vertically mounted high-mounted ultrasonic sensor and a low-mounted ultrasonic sensor, with the first landing point of the detection range of the high-mounted ultrasonic sensor coinciding with the second landing point of the detection range of the low-mounted ultrasonic sensor; and determining target obstacles around the self-moving device and their target location information based on the obstacle detection information. In this application, by installing an ultrasonic sensor group on the self-moving device, obstacles within the radar perception blind zone can be detected through the ultrasonic sensor group, thereby ensuring the movement safety of the self-moving device.

The invention discloses a fixed-wing unmanned aerial vehicle landing navigation method, device and equipment and a medium, and relates to the technical field of unmanned aerial vehicle control, and the method comprises the following steps: collecting an image of an unmanned aerial vehicle in a landing stage, and determining a runway area from the image; gray processing is carried out on the image, candidate line segments in the gray image are extracted by using a region growing algorithm, and a horizon line in the forward view angle image is determined; screening the candidate line segments to obtain a screened runway line set; performing runway line detection on the runway area to obtain a corresponding runway line detection result, and performing comparison verification on line segments in the screened runway line set and the runway line detection result to determine a target runway line; and carrying out landing navigation based on the target runway line, the horizon line and the camera internal reference pair of the preset camera. False detection caused by shadows, interference lines and the like is effectively eliminated through screening based on geometric constraints and a consistency verification means among detection results of different algorithms, so that the reliability of landing of the unmanned aerial vehicle is ensured.

The application discloses a farmland abnormal map spot remote sensing identification method and device, equipment and a storage medium, relates to the technical field of remote sensing identification, and the farmland abnormal map spot remote sensing identification method comprises the following steps: acquiring multispectral remote sensing images and farmland vector boundary data of a target area; multi-dimensional feature screening is carried out on the multispectral remote sensing images to obtain target features for representing farmland abnormal targets; feature enhancement processing is carried out on the multispectral remote sensing images according to the target features to generate feature enhancement images; the feature enhancement images are input into a deep learning semantic segmentation model to obtain preliminary abnormal map spot extraction results; and spatial constraint processing is carried out on the preliminary abnormal map spot extraction results based on the farmland vector boundary data to obtain farmland abnormal map spots. The application can improve the identification precision of farmland abnormal map spots.

This invention discloses a method and system for compliance review of tourism planning based on a dynamic weighted resource coupling model. The method includes: constructing a general standard and a dedicated feature knowledge base; reviewing the integrity of electronic documents of the tourism plan to be reviewed using ontology mapping; extracting carrying capacity indicators and substituting them into the dynamic weighted resource coupling model to calculate theoretical facility requirements and compare them with actual designs; verifying the consistency of textual and graphical parameters through multimodal analysis technology; and finally generating a tiered early warning report. This invention solves the problem of insufficient scientific basis for resource allocation in tourism planning through a quantitative model, and achieves fully automated intelligent review from text to drawings, with continuous evolution capabilities combined with an expert feedback mechanism.

The present invention discloses a method for detecting faults of electric bicycles, and relates to the technical field of electric bicycle detection, and comprises the following steps: step one, establishing a triangular linkage system among a driving input module, a power output module and a driving module; step two: when the driving input module, the power output module and the driving module are working, three corresponding sets of judgment logic are formed; step three: when a set of modules is judged to be faulty in a single judgment logic, the remaining judgment logics are started, a self-checking process is performed, and the self-checking results are obtained; step four: the self-checking results are analyzed to obtain a fault analysis. The present invention realizes a comprehensive detection of the running state of the electric bicycle through the triangular linkage system of the driving input module, the power output module and the driving module and multiple sets of judgment logics, and the interaction and feedback mechanism between the modules ensures the accuracy of the detection, and through the data comparison between the modules, it can be more accurately determined whether there is a fault.