10results about How to "Various locations" patented technology

The invention discloses a defense method oriented to adversarial patch attacks, and the defense method oriented to adversarial patch attacks is realized by designing an adversarial mask generation method and a classification-reconstruction joint training method by utilizing the characteristic that a mask reconstruction task has good feature extraction capability. According to the defense method facing the adversarial patch attack, an image recognition model is divided into a backbone network and a full connection layer which are respectively used for feature extraction and classification prediction, and after a reconstruction model is connected to the backbone network, the reconstruction model is used for receiving image features extracted by the backbone network and reconstructing an image. The connected image recognition model backbone network, the full connection layer and the reconstruction model are generally called as training models. The connection mode does not need to modify the structure of an original image recognition model, only optimizes the weight parameters of the original model, is suitable for deep learning visual models of any architecture, is suitable for models in any training state, and does not need to forcibly train from the beginning.

The invention relates to the technical field of power distribution cabinets, in particular to a power distribution cabinet dehumidification assembly and a power distribution cabinet, the power distribution cabinet dehumidification assembly comprises a vertical support and a dehumidification mechanism, and the dehumidification mechanism comprises a horizontal cylinder, a dehumidification fan, a heating branch chain, a control branch chain and a driving branch chain; the problems that in the prior art, in the dehumidification process of the interior of the power distribution cabinet, the dehumidification range is limited, comprehensive dehumidification cannot be achieved, dead corners exist, and the dehumidification effect is affected can be solved; heated air in the power distribution cabinet cannot be exhausted, so that rising water vapor is condensed on the inner top wall of the power distribution cabinet to form water, and short circuit and potential safety hazards are easily caused; air is blown into the power distribution cabinet for dehumidification in an internal circulation mode, and air in the power distribution cabinet is sucked into the horizontal cylinder for drying, so that the dryness in the power distribution cabinet is ensured, heated air is prevented from being condensed into water without being discharged, and the dehumidification effect can be enhanced; by changing the air blowing direction of the dehumidification fan, comprehensive dehumidification can be achieved.

The present invention belongs to the technical field of lithium battery detection, and specifically is a device for detecting the impact resistance of a cylindrical lithium battery, comprising an impact testing machine, wherein two lifting guide rails symmetrically arranged in a group and a base fixedly installed at the bottom are fixedly installed inside the impact testing machine, and a top plate is slidably connected to the surface of the lifting guide rails; the present invention starts a servo motor, uses a lead screw to drive a sleeve plate and an impact block to move to a suitable position, and then cuts off the power to the electromagnet in the sleeve plate. At this time, the electromagnet at the bottom of the sleeve plate will no longer adsorb the impact block, so that the impact block falls with the insertion block in the direction of the cylindrical lithium battery, so that different parts of the battery can be subjected to impact detection at one time, which greatly improves the detection efficiency and makes the detection more complete. At the same time, the position of the impact block can be conveniently adjusted for cylindrical lithium batteries of different lengths, so as to achieve the purpose of accurate detection.

The present application discloses a license plate recognition method, an electronic device and a storage medium. The license plate recognition method includes: obtaining at least one initial image feature, each initial image feature is obtained by extracting the features of the image to be detected collected from the license plate; based on each initial image feature, determining the character offset information about the license plate in each initial image feature and the position information of the area of ​​interest, the area of ​​interest is the area where the license plate is located in the image to be detected, and each character offset information is used to represent the position relationship between at least part of the characters in the image to be detected and the preset position in the license plate; based on each character offset information and the position information of each area of ​​interest, determining the license plate recognition result of the image to be detected, the license plate recognition result includes at least one of the following: the type of the license plate, the area where the license plate is located in the image to be detected, and the character information in the license plate. The above scheme can improve the accuracy of license plate recognition.

The invention belongs to the technical field of detection devices, and discloses a magnetic shoe interior detection device and method, and the device comprises a detection frame, a transmission structure is mounted in the detection frame, a positioning assembly is arranged above the transmission structure, a mounting frame is fixedly mounted at the top end of the detection frame, and a detection structure is mounted at the bottom end of the mounting frame. The detection structure is arranged above the detection frame, a first rotating shaft is mounted in the transmission structure, a first gear is fixedly mounted on the outer wall of the first rotating shaft, a second gear is in meshed connection with one side of the first gear, a second rotating shaft is fixedly mounted in the second gear, and a motor is arranged at one end of the second rotating shaft; when the magnetic shoe is detected, the magnetic shoe is firstly placed at the top end of the transmission structure, then the transmission structure works to transmit the magnetic shoe to one side of the detection structure, the magnetic shoe is stopped after moving to the position below the detection structure, and finally the static magnetic shoe is detected through the detection structure, so that the accuracy of a magnetic shoe detection result is improved.

The invention discloses a wire end connector and a connector module, and is applied to the field of electronic equipment, the wire end connector comprises a wire end connector body, and the wire end connector body is provided with an input connecting end used for installing a hard disk; the output connecting end is used for connecting a cable; and the conductive connecting end is used for being detachably connected with a board end connector body, and the orientation of the input connecting end is perpendicular to the orientation of the conductive connecting end. The wire end connector disclosed by the invention can be connected with equipment such as a hard disk, a cable and a board end connector body at the same time, can select a proper model according to the equipment such as the hard disk while meeting the flexibility, is wide in application range, and can realize an expansion function.

The application provides a construction worker unsafe behavior intelligent identification method and early warning system, the method comprises the following steps: obtaining a to-be-detected image of a construction site; inputting the to-be-detected image into a construction worker unsafe behavior identification model to obtain an unsafe behavior identification result of a construction worker; wherein the unsafe behavior at least includes an unworn state, a wrong wearing state or a blocked state of personal protective equipment; and the construction worker unsafe behavior identification model is a neural network model for target detection. The detection accuracy is significantly improved, and the fine-grained identification capability is enhanced. Since a learnable position coding component is introduced into the backbone network, the model can explicitly model the spatial position relationship of the feature map, enhance the spatial position perception capability of the personnel (especially small targets and blocked targets) in the construction scene under complex background interference, solve the problem that the traditional convolutional network is not sensitive to absolute position information, and improve the feature extraction accuracy in a complex background.

The application relates to a target detection method, device and computer equipment under a complex traffic scene. The method comprises the following steps: determining a detection result of a to-be-detected image based on a traffic target detection model; a mixed encoder layer of the traffic target detection model comprises an FDPN-P2 network; the detection result of the to-be-detected image is determined based on the traffic target detection model, which comprises the following steps: inputting original features obtained by processing the to-be-detected image into a plurality of residual blocks of a backbone network to obtain scale features,, and ; based on a feature fusion unit and a RepC3 unit of the FDPN-P2 network, the scale features and and the scale features after self-attention processing are subjected to fusion processing to obtain multi-scale features, and, and the multi-scale features after preprocessing and the scale features are fused to obtain multi-scale features; a plurality of Tokens are obtained based on the multi-scale features,, and, and a detection result of the to-be-detected image is obtained based on the plurality of Tokens. The method can improve the reliability of target detection under a complex traffic scene.

The invention discloses a grinding device capable of improving the smoothness of an impeller flow channel, comprising a sand box, corundum, an impeller, a sleeve shaft, a bearing box and a variable frequency speed regulating motor, wherein the sand box is a hollow structure, and one end of the variable frequency speed regulating motor is penetrated by a main shaft with a cylindrical structure. The invention has the following beneficial effects: corundum is continuously thrown out by impeller blades through the impeller flow channel, and the throwing direction is consistent with the direction of the impeller flow channel profile. Since corundum has fine grains and high hardness, the impeller flow channel is polished smooth by flushing the impeller flow channel. The invention can easily and quickly remove casting burrs and sand inclusions in the impeller flow channel, and can greatly improve the smoothness in the impeller flow channel. A large amount of corundum fine grains are flushed on the inner wall of the impeller flow channel, while ensuring that the device is more efficient in grinding and removing burrs from the impeller flow channel, the device also covers the position and angle of the impeller flow channel grinding work more comprehensively and removes burrs more thoroughly.

The invention discloses a remote sensing image building detection method, which belongs to the technical field of deep learning, and designs an improved YOLOv7 network architecture on the basis of a YOLOv7 algorithm in combination with a coordinate attention sensing module, a multi-scale residual module and a feature-refined dense connection feature pyramid module. Through the innovative design, the interference of a complex background is further reduced, and the problems caused by the size change of a building and unobvious characteristics of a small building are effectively relieved, so that the robustness and precision of target detection are improved.