The present disclosure describes a method, an apparatus, and a storage medium for recognizing an obstacle. The method includes acquiring, by a device, point cloud data obtained by scanning surroundings of a target vehicle by a sensor in the target vehicle. The device includes a memory storing instructions and a processor in communication with the memory. The method further includes converting, by the device, the point cloud data into a first image used for showing the surroundings; and recognizing, by the device, from the first image, a first object in the surroundings as an obstacle through a first neural network model.

An identification module is disclosed for use in a key making machine. The identification module may have a key receiving assembly configured to receive only a shank of an existing key. The identification module may also have a tip guide, configured to receive a tip of the shank of the existing key. The tip guide may have a slot that exposes a tip end of the shank. The identification module may also have an imaging assembly configured to capture an image of the tip end through the slot.

An identification module is disclosed for use in a key making machine. The identification module may have a key receiving assembly configured to receive only a shank of an existing key. The identification module may also have a tip guide, configured to receive a tip of the shank of the existing key. The tip guide may have a slot that exposes a tip end of the shank. The identification module may also have an imaging assembly configured to capture an image of the tip end through the slot.

An artificial neural network-based method for selecting a surface type of an object includes receiving at least one object image, performing surface type identification on each of the at least one object image by using a first predictive model to categorize the object image to one of a first normal group and a first abnormal group, and performing surface type identification on each output image in the first normal group by using a second predictive model to categorize the output image to one of a second normal group and a second abnormal group.

An artificial neural network-based method for selecting a surface type of an object includes receiving at least one object image, performing surface type identification on each of the at least one object image by using a first predictive model to categorize the object image to one of a first normal group and a first abnormal group, and performing surface type identification on each output image in the first normal group by using a second predictive model to categorize the output image to one of a second normal group and a second abnormal group.

A facial tracking device including an illuminator and a photon detector. The illuminator configured to project a light toward a first portion of a head of a user and the photon detector configured to detect light reflected from a second portion of the head of the user. The facial tracking device further including a processor that is connected to the illuminator and photon detector and configured to cause the illuminator to project the light toward the head of the user, receive information from the photon detector, and determine a facial expression of the user.

A facial tracking device including an illuminator and a photon detector. The illuminator configured to project a light toward a first portion of a head of a user and the photon detector configured to detect light reflected from a second portion of the head of the user. The facial tracking device further including a processor that is connected to the illuminator and photon detector and configured to cause the illuminator to project the light toward the head of the user, receive information from the photon detector, and determine a facial expression of the user.

An image processing system includes a detection unit, an identification unit, and a feature extraction unit. The detection unit detects, from a first image in which a target person is captured, a candidate region being an image region estimated to represent an eye of the target person, based on a first evaluation value. The identification unit identifies, from the detected candidate region, an eye region being an image region that represents the eye, based on a second evaluation value. The feature extraction unit extracts a feature value of the identified eye region. The first evaluation value indicates a likelihood of the eye, and is calculated for an image region being set based on the first image. The second evaluation value indicates a likelihood of the eye, and is calculated for an image region being set based on the detected candidate region.

An apparatus for detecting a specular surface in a scene is provided. The apparatus includes an illumination device configured to emit polarized light towards the scene. The apparatus further includes an imaging system configured to capture a first image of the scene based on light emanating from the scene. The light emanating from the scene includes one or more reflection of the emitted polarized light. The imaging system is further configured to capture a second image of the scene based on filtered light. The apparatus further includes a polarization filter configured to generate the filtered light by filtering the light emanating from the scene. The apparatus further includes processing circuitry configured to determine presence of the specular surface in the scene based on a comparison of the first image and the second image.

A light distribution controller includes a computation unit that extracts an index value from a predetermined region to be processed within an image, a pattern determiner that determines a light distribution pattern such that the index value approaches a maximum value, and a lamp controller that controls a light distribution variable lamp so as to form the light distribution pattern. The index value is at least one of a mean intensity of HOG feature values of a plurality of pixels of the region to be processed or an edge gray level proportion indicating, in a local binary pattern (LBP) histogram of the plurality of pixels, a proportion of the number of pixels belonging to a gray level of an edge portion to a total number of the plurality of pixels.