Abstract: The present disclosure discloses an adaptive high-precision compression method and system based on a convolutional neural network model, and belongs to the fields of artificial intelligence, computer vision, and image processing. According to the method of the present disclosure, coarse-grained pruning is performed on a neural network model by using a differential evolution algorithm first, and the coarse-grained space is quickly searched through an entropy importance criterion and an objective function with good guidance to obtain a near-optimal neural network structure. Then fine-grained search space is built on the basis of an optimal individual obtained from the coarse-grained search, and fine-grained pruning is performed on the neural network model by a differential evolution algorithm to obtain a network model with an optimal structure. Finally, the performance of the optimal model is restored by using a multi-teacher multi-step knowledge distillation network to reach the precision of an original model.

Abstract: Disclosed is a snow shoveling and snow discharging assembly of a snow sweeping robot. The snow shoveling and snow discharging assembly comprises a snow stirring structure, a snow feeding structure and a snow raising structure. Stirring cutters in the snow stirring mechanism are driven by a stirring cutter shaft to stir bottom area snow into a snow feeding pipe of the snow feeding structure, an air blower in the snow feeding structure blows accumulated snow into a snow raising pipe of the snow raising structure through a connecting pipe, the snow raising pipe can be driven by a steering motor to rotate by a certain angle to control the snow raising direction, and a second electric push rod acts to control pitching of the guide part, so that the height of snow during snow raising is controlled.

Abstract: The present disclosure provides a decentralized active equalization method for a cascaded lithium-ion battery pack. The method includes: connecting each battery cell in the cascaded lithium-ion battery pack to a direct current (DC) bus through an equalizer respectively, where each equalizer includes an independent controller, a sampling circuit, a power supply circuit, a drive circuit, and a main circuit; connecting an input terminal of the main circuit to a corresponding battery cell, and connecting an output terminal of the main circuit to the DC bus. The present disclosure solves the technical problem that an existing cascaded lithium-ion battery pack equalization method cannot achieve equalization when a centralized controller failure or a communication failure occurs, improves the reliability of the equalization method, can make the equalizer work at high efficiency by configuring parameters of C, K, and R, and speeds up the equalization or improves the equalization accuracy.

Abstract: Embodiments of the present disclosure provide a quantitative method and system for attention based on a line-of-sight estimation neural network, which improves the stability and training efficiency of the line-of-sight estimation neural network. A few-sample learning method is applied to training of the line-of-sight estimation neural network, which improves generalization performance of the line-of-sight estimation neural network. A nonlinear division method for small intervals of angles of the line of sight is provided, which reduces an estimation error of the line-of-sight estimation neural network. Eye opening and closing detection is added to avoid the line-of-sight estimation error caused by an eye closing state. A method for solving a landing point of the line of sight is provided, which has high environmental adaptability and can be quickly used in actual deployment.

Abstract: The present disclosure provides a real-time vehicle overload detection method based on a convolutional neural network (CNN). The present disclosure detects a road driving vehicle in real time with a CNN method and a you only look once (YOLO)-V3 detection algorithm, detects the number of wheels to obtain the number of axles, detects a relative wheelbase, compares the number of axles and the relative wheelbase with a national vehicle load standard to obtain a maximum load of the vehicle, and compares the maximum load with an actual load measured by a piezoelectric sensor under the vehicle, thereby implementing real-time vehicle overload detection. The present disclosure has desirable real-time detection, can implement no-parking vehicle overload detection on the road, and avoids potential traffic congestions and road traffic accidents.

Abstract: The present disclosure provides a neural network-based method for calibration and localization of an indoor inspection robot. The method includes the following steps: presetting positions for N label signal sources capable of transmitting radio frequency (RF) signals; computing an actual path of the robot according to numbers of signal labels received at different moments; computing positional information moved by the robot at a tth moment, and computing a predicted path at the tth moment according to the positional information; establishing an odometry error model with the neural network and training the odometry error model; and inputting the predicted path at the tth moment to a well-trained odometry error model to obtain an optimized predicted path. The present disclosure maximizes the localization accuracy for the indoor robot by minimizing the error of the odometer with the odometry calibration method.

Abstract: The present disclosure relates to a method for recognizing facial expressions based on adversarial elimination. First, a facial expression recognition network is built based on a deep convolutional neural network. On a natural facial expression data set, the facial expression recognition network is trained through a loss function to make facial expression features easier to distinguish. Then some key features of input images are actively eliminated by using an improved confrontation elimination method to generate a new data set to train new networks with different weight distributions and feature extraction capabilities, forcing the network to perform expression classification discrimination based on more features, which reduces the influence of interference factors such as occlusion on the network recognition accuracy rate, and improving the robustness of the facial expression recognition network.

Abstract: Disclosed is a snow shoveling and snow discharging assembly of a snow sweeping robot. The snow shoveling and snow discharging assembly comprises a snow stirring structure, a snow feeding structure and a snow raising structure. Stirring cutters in the snow stirring mechanism are driven by a stirring cutter shaft to stir bottom area snow into a snow feeding pipe of the snow feeding structure, an air blower in the snow feeding structure blows accumulated snow into a snow raising pipe of the snow raising structure through a connecting pipe, the snow raising pipe can be driven by a steering motor to rotate by a certain angle to control the snow raising direction, and a second electric push rod acts to control pitching of the guide part, so that the height of snow during snow raising is controlled.

Abstract: The present disclosure discloses an adaptive high-precision compression method and system based on a convolutional neural network model, and belongs to the fields of artificial intelligence, computer vision, and image processing. According to the method of the present disclosure, coarse-grained pruning is performed on a neural network model by using a differential evolution algorithm first, and the coarse-grained space is quickly searched through an entropy importance criterion and an objective function with good guidance to obtain a near-optimal neural network structure. Then fine-grained search space is built on the basis of an optimal individual obtained from the coarse-grained search, and fine-grained pruning is performed on the neural network model by a differential evolution algorithm to obtain a network model with an optimal structure. Finally, the performance of the optimal model is restored by using a multi-teacher multi-step knowledge distillation network to reach the precision of an original model.

Abstract: The present disclosure relates to a method for recognizing facial expressions based on adversarial elimination. First, a facial expression recognition network is built based on a deep convolutional neural network. On a natural facial expression data set, the facial expression recognition network is trained through a loss function to make facial expression features easier to distinguish. Then some key features of input images are actively eliminated by using an improved confrontation elimination method to generate a new data set to train new networks with different weight distributions and feature extraction capabilities, forcing the network to perform expression classification discrimination based on more features, which reduces the influence of interference factors such as occlusion on the network recognition accuracy rate, and improving the robustness of the facial expression recognition network.

Abstract: The present disclosure provides a real-time vehicle overload detection method based on a convolutional neural network (CNN). The present disclosure detects a road driving vehicle in real time with a CNN method and a you only look once (YOLO)-V3 detection algorithm, detects the number of wheels to obtain the number of axles, detects a relative wheelbase, compares the number of axles and the relative wheelbase with a national vehicle load standard to obtain a maximum load of the vehicle, and compares the maximum load with an actual load measured by a piezoelectric sensor under the vehicle, thereby implementing real-time vehicle overload detection. The present disclosure has desirable real-time detection, can implement no-parking vehicle overload detection on the road, and avoids potential traffic congestions and road traffic accidents.

Abstract: An unmanned aerial vehicle navigation map construction system based on three-dimensional image reconstruction technology comprises an unmanned aerial vehicle, a data acquiring component and a three-dimensional navigation map construction system, wherein the three-dimensional navigation map construction system comprises an image set input system, a feature point extraction system, a sparse three-dimensional point cloud reconstruction system, a dense three-dimensional point cloud reconstruction system, a point cloud model optimization system and a three-dimensional navigation map reconstruction system. A scene image set is input into the three-dimensional navigation map construction system, feature point detection is carried out on all images, a sparse point cloud model of the scene and a dense point cloud model of the scene are reconstructed, the model is optimized by removing a miscellaneous point and reconstructing the surface, and a three-dimensional navigation map of the scene is reconstructed.