Abstract: A method and device for 3D object detection. The method comprises the steps of: generating one or more fused images(201) based on a pair of images(530), the pair of images (530) including a left view image(101,710a) and a right view image(102,710b); extracting one or more fused features from the fused images(201) by a single backbone network(210) of a share network with feature unmixing (SNFU,540); unmixing the fused features to a left view-aware feature and a right view-aware feature by a feature unmixing sub-network(220) of the SNFU(540); predicting the 3D object based on the left view-aware feature and the right view-aware feature; and determining spatial features of the predicted 3D object. The proposed method and device can reduce the computation complexity to the level of that in monocular based 2D object detection so as to improve the computation efficiency while keeping high precision.

Abstract: Method, apparatus, computer program product and computer readable medium are disclosed for biometric recognition. The method may comprise obtaining first biometric data, wherein the first biometric data is captured in first condition or non-first condition; determining, by a discriminative network of generative adversarial networks, whether the first biometric data is captured in the first condition or non-first condition; in response to a determination that the first biometric data is captured in the non-first condition, inputting the first biometric data to a generative network of the generative adversarial networks to generate a second biometric data; obtaining a matching result by inputting the first biometric data in response to a determination that the first biometric data is captured in the first condition or the second biometric data to a matching network; and determining a recognition result based on the matching result.

Abstract: The present application provides a method and apparatus for detecting an ECG signal and an electronic device. According to an example of the method, an ECG signal with a set time length is segmented to obtain a first set number of single heartbeats; feature data corresponding to each of the first set number of single heartbeats is determined to obtain a first set number of feature data; and a pathological category of the ECG signal with the set time length is determined based on the ECG signal with the set time length and the first set number of feature data.

Abstract: Methods, apparatuses and computer program products for object detection. A method comprises extracting a generic feature of an image characterizing one or more general properties of the image(310); identifying one or more regions of interest (ROIs) (320); generating scale information on one or more objects in the image based on the generic feature and one or more candidate scales(330); generating one or more scale-specific features of the image based on the scale information(340); and detecting the one or more objects in the image based on the identified one or more ROIs and the one or more scale-specific features(350).

Abstract: A method, apparatus and computer program product for removing weather elements, such as rain, from images are provided. In this regard, imagery from self-driving cars or video surveillance may be blurred by weather elements. The weather elements may be removed by utilizing pure weather element images. The pure weather element images may be processed by machine learning techniques to model pure weather element data, and generate a pure weather element dictionary. Imagery including both weather elements and background scenery may then be processed accordingly, and the weather elements removed based on the learned pure weather element dictionary. Resulting weather-free images may then be generated.

Abstract: Method, apparatus, computer program product and computer readable medium are disclosed for biometric recognition. The method may comprise obtaining first biometric data, wherein the first biometric data is captured in first condition or non-first condition; determining, by a discriminative network of generative adversarial networks, whether the first biometric data is captured in the first condition or non-first condition; in response to a determination that the first biometric data is captured in the non-first condition, inputting the first biometric data to a generative network of the generative adversarial networks to generate a second biometric data; obtaining a matching result by inputting the first biometric data in response to a determination that the first biometric data is captured in the first condition or the second biometric data to a matching network; and determining a recognition result based on the matching result.

Abstract: The present application provides a method and apparatus for detecting an ECG signal and an electronic device. According to an example of the method, an ECG signal with a set time length is segmented to obtain a first set number of single heartbeats; feature data corresponding to each of the first set number of single heartbeats is determined to obtain a first set number of feature data; and a pathological category of the ECG signal with the set time length is determined based on the ECG signal with the set time length and the first set number of feature data.

Abstract: A method, apparatus and computer program product for removing weather elements, such as rain, from images are provided. In this regard, imagery from self-driving cars or video surveillance may be blurred by weather elements. The weather elements may be removed by utilizing pure weather element images. The pure weather element images may be processed by machine learning techniques to model pure weather element data, and generate a pure weather element dictionary. Imagery including both weather elements and background scenery may then be processed accordingly, and the weather elements removed based on the learned pure weather element dictionary. Resulting weather-free images may then be generated.

Abstract: In accordance with an example embodiment of the present invention, a method comprising: obtaining a plurality of training samples; employing a set of activation functions on a plurality of layers of a deep neural network, wherein the set of activation functions varies with the plurality of layers; and applying the activation functions on the plurality of training samples.

Abstract: In accordance with an example embodiment of the present invention, a method comprising: obtaining a plurality of training cases; initializing a filter corresponding to each convolutional layer in a convolutional neural network, wherein the convolutional neural network comprises at least one convolutional layer; applying a squashing function on the filter; computing convolutions of patches from the plurality of training images and the filter which has applied the squashing function; and obtaining parameters of the squashing function and parameters of the filter based on the computed convolutions.