Abstract: Videos captured in low light conditions can be processed in order to identify an activity being performed in the video. The processing may use both the video and audio streams for identifying the activity in the low light video. The video portion is processed to generate a darkness-aware feature which may be used to modulate the features generated from the audio and video features. The audio features may be used to generate a video attention feature and the video features may be used to generate an audio attention feature. The audio and video attention features may also be used in modulating the audio video features. The modulated audio and video features may be used to predict an activity occurring in the video.

Abstract: Videos captured in low light conditions can be processed in order to identify an activity being performed in the video. The processing may use both the video and audio streams for identifying the activity in the low light video. The video portion is processed to generate a darkness-aware feature which may be used to modulate the features generated from the audio and video features. The audio features may be used to generate a video attention feature and the video features may be used to generate an audio attention feature. The audio and video attention features may also be used in modulating the audio video features. The modulated audio and video features may be used to predict an activity occurring in the video.

Abstract: Methods, systems, and techniques for kernel continuing learning. A dataset is obtained that corresponds to a classification task. Feature extraction is performed on the dataset using an artificial neural network. A kernel is constructed using features extracted during that feature extraction for use in performing the classification task. More particularly, during training, a coreset dataset corresponding to the classification task is saved; and during subsequent inference, the coreset dataset is retrieved and used to construct a task-specific kernel for classification.

Abstract: The present disclosure provides a dense crowd counting method and an apparatus, including: acquiring an image to be detected, where the image to be detected includes images of people; feeding the image to be detected into a convolutional neural network model to obtain a crowd density map of the image to be detected; and determining the number of the images of people in the image to be detected according to the crowd density map. Feature information of an image to be detected may be fully extracted through the above mentioned process, thereby realizing a better performance of crowd counting and density estimation, providing great convenience for subsequent security monitoring, crowd control and other applications.

Abstract: The present disclosure provides a target tracking method and device oriented to airborne-based monitoring scenarios. The method includes: obtaining a video to-be-tracked of the target object in real time; extracting a first frame and a second frame; trimming and capturing the first frame to derive an image for first interest region; trimming and capturing the second frame to derive an image for target template and an image for second interest region; inputting the image for target template and the image for first interest region into an appearance tracker network to derive an appearance tracking position; inputting the image for first interest region and the image for second interest region into a motion tracker network to derive a motion tracking position; and finally inputting the appearance tracking position and the motion tracking position into a deep integration network to derive a final tracking position.

Abstract: The present disclosure provides a dense crowd counting method and an apparatus, including: acquiring an image to be detected, where the image to be detected includes images of people; feeding the image to be detected into a convolutional neural network model to obtain a crowd density map of the image to be detected; and determining the number of the images of people in the image to be detected according to the crowd density map. Feature information of an image to be detected may be fully extracted through the above mentioned process, thereby realizing a better performance of crowd counting and density estimation, providing great convenience for subsequent security monitoring, crowd control and other applications.

Abstract: The present disclosure provides a target tracking method and device oriented to airborne-based monitoring scenarios. The method includes: obtaining a video to-be-tracked of the target object in real time; extracting a first frame and a second frame; trimming and capturing the first frame to derive an image for first interest region; trimming and capturing the second frame to derive an image for target template and an image for second interest region; inputting the image for target template and the image for first interest region into an appearance tracker network to derive an appearance tracking position; inputting the image for first interest region and the image for second interest region into a motion tracker network to derive a motion tracking position; and finally inputting the appearance tracking position and the motion tracking position into a deep integration network to derive a final tracking position.