Existing single-stage target detection algorithms are insensitive to nodule detection in lung nodule detection， multiple up-samplings during feature extraction by Convolutional Neural Network （CNN） has difficult feature extraction and poor detection effect， and the existing pulmonary nodule detection algorithm models are complex and not conductive to practical application employment and implementation. To address the above problems， a real-time pulmonary nodule detection algorithm combining attention mechanism and multipath fusion was proposed， based on which the up-sampling algorithm was improved to effectively increase the detection accuracy of lung nodules and speed of model inference， the model size was small and easy to deploy. Firstly， the hybrid attention mechanism of channel and space was fused in the backbone network part of feature extraction. Secondly， the sampling algorithm was improved to enhance the quality of generated feature maps. Finally， the channels were established between different paths in the enhanced feature extraction network part to achieve the fusion of deep and shallow features， so the semantic and location information at different scales was fused. Experimental results on LUNA16 dataset show that， compared to the original YOLOv5s algorithm， the proposed algorithm achieves an improvement of 9.5， 6.9， and 8.7 percentage points in precision， recall， and average precision， respectively， with a frame rate of 131.6 frames/s， and a model weight file of only 14.2 MB， demonstrating that the proposed algorithm can detect lung nodules in real time with much higher accuracy than existing single-stage detection algorithms such as YOLOv3 and YOLOv8.

To solve the high consumption problem of fast video Shot Boundary Detection (SBD), an improved Shot Boundary Detection (SBD) method was proposed based on video pre-processing. Adaptive threshold was taken to select the candidate segments which may contain shot boundaries, the beginning frame was detected by comparing the similarity value between the first frame and the rest frames in the candidate segments, and then cut was detected immediately. Gradual transition would be detected if no cut was detected. Candidate segments had to be adjusted to ensure the whole transition was located in the same segment. Ending frame was confirmed by comparing the similarity value between the beginning frame and the rest frames in segment. Experimental results demonstrate that the proposed algorithm achieves an accuracy above 90% and the time cost is reduced 15.6%-30.2% compared with inverted triangle pattern matching method. The proposed algorithm satisfies the need of accuracy and improves detection speed compared with the traditional methods which need detection for both cut and gradual transition.

To build a common user model platform which can share data and provide more comprehensive and accurate user information to all sites accessed to the platform, the platform provided data interfaces and algorithm interfaces to interact with the third-party sites, focused on how to solve the conflict of user data from different data sources in order to form a unified user model, finally achieved sharing of algorithms, models and data. The experimental results show that it is more accurate and comprehensive to build user model on this platform.