Described herein are systems, methods, and non-transitory computer readable media for using 3D point cloud data such as that captured by a LiDAR as ground truth data for training an instance segmentation deep learning model. 3D point cloud data captured by a LiDAR can be projected on a 2D image captured by a camera and provided as input to a 2D instance segmentation model. 2D sparse instance segmentation masks may be generated from the 2D image with the projected 3D data points. These 2D sparse masks can be used to propagate loss during training of the model. Generation and use of the 2D image data with the projected 3D data points as well as the 2D sparse instance segmentation masks for training the instance segmentation model obviates the need to generate and use actual instance segmentation data for training, thereby providing an improved technique for training an instance segmentation model.

A semantic segmentation method and system for a high-resolution remote sensing image based on random blocks. In the semantic segmentation method, the high-resolution remote sensing image is divided into random blocks, and semantic segmentation is performed for each individual random block separately, thus avoiding overflow of GPU memory during semantic segmentation of the high-resolution remote sensing image. In addition, feature data in random blocks neighboring each random block incorporated into the process of semantic segmentation, overcoming the technical shortcoming that the existing segmentation method for the remote sensing image weakens the correlation within the image. Moreover, in the semantic segmentation method, semantic segmentation is separately performed on mono-spectral feature data in each band of the high-resolution remote sensing image, thus enhancing the accuracy of sematic segmentation of the high-resolution remote sensing image.

A method for detecting the presence of on-body concealed objects includes receiving a visible-domain camera image for a scene, determining, using the visible-domain camera image, a region of interest where a subject is present, receiving an infrared-domain camera image and a millimeter-wave (mmwave) radar image that each cover the region of interest, determining emissivity information for the region of interest using the infrared-domain camera image, determining reflectivity information for the region of interest using the mmwave radar image and determining a concealed object classification for the subject based on the emissivity information and the reflectivity information. A corresponding system and computer program product for executing the above method are also disclosed herein.

A hearing aid system for selective modification of background noises may include at least one processor. The processor may be programmed to receive a plurality of images from an environment of a user captured by a wearable camera during a time period and receive an audio signal representative of sounds acquired by a wearable microphone during the time period. The processor may determine that at least one of the sounds was generated by a sound-emanating object in the environment of the user, but outside of a field of view of the wearable camera and retrieve from a database information associated with the sound-emanating object. Based on the retrieved information, the processor may cause selective conditioning of audio signals acquired by the wearable microphone during the time period and cause transmission of the conditioned audio signals to a hearing interface device configured to provide sounds to an ear of the user.

A unified imaging device used for detecting and classifying objects in a scene including motion and micro-vibrations by receiving a plurality of images of the scene captured by an imaging sensor of the unified imaging device comprising a light source adapted to project on the scene a predefined structured light pattern constructed of a plurality of diffused light elements, classifying object(s) present in the scene by visually analyzing the image(s), extracting depth data of the object(s) by analyzing position of diffused light element(s) reflected from the object(s), identifying micro-vibration(s) of the object(s) by analyzing a change in a speckle pattern of the reflected diffused light element(s) in at least some consecutive images and outputting the classification, the depth data and data of the one or more micro-vibrations which are derived from the analyses of images captured by the imaging sensor and are hence inherently registered in a common coordinate system.

A processor-implemented liveness detection method includes: obtaining an initial image using a dual pixel sensor; obtaining a left image and a right image from the initial image; and detecting liveness of an object included in the initial image using the left image and the right image.

A main video sequence of a live action scene is captured along with ancillary device data to provide corresponding volumetric information about the scene. The volumetric data can then be used to visually remove or replace objects in the main video sequence. A removed object is replaced by the view that would have been captured by the main video sequence had the removed object not been present in the live action scene at the time of capturing.

Distance measurements are received from one or more distance measurement sensors, which may be coupled to a vehicle. A three-dimensional (3D) point cloud are generated based on the distance measurements. In some cases, 3D point clouds corresponding to distance measurements from different distance measurement sensors may be combined into one 3D point cloud. A voxelized model is generated based on the 3D point cloud. An object may be detected within the voxelized model, and in some cases may be classified by object type. If the distance measurement sensors are coupled to a vehicle, the vehicle may avoid the detected object.

One or more multi-stage optimization iterations are performed with respect to a compression algorithm. A given iteration comprises a first stage in which hyper-parameters of a perceptual quality algorithm are tuned independently of the compression algorithm. A second stage of the iteration comprises tuning hyper-parameters of the compression algorithm using a set of perceptual quality scores generated by the tuned perceptual quality algorithm. The final stage of the iteration comprises performing a compression quality evaluation test on the tuned compression algorithm.

The exemplary embodiments disclose a system and method, a computer program product, and a computer system for generating an agriculture map of a region of land. The exemplary embodiments may include collecting agricultural data of one or more sub-regions of the region of land, wherein the agricultural data includes classified data and unclassified data, extracting one or more features from the collected classified agricultural data, training one or more models based on the extracted one or more features, and generating an agricultural map of the region of land based on applying the one or more models to the collected unclassified agricultural data.