Abstract: A processing method of face recognition includes steps of: extracting embedding feature information from a face image; outputting a recognition result of face recognition according to the embedding feature information, wherein the recognition result includes a recognized name and embedding feature distance information; determining whether the recognized name is in a list or not; if the recognized name is in the list, performing a removal checking step for determining whether to remove the recognition result based on the embedding feature distance information; if determining that the recognition result is not to be removed, displaying the recognized name; if determining that the recognition result is to be removed, displaying a negative prompt; and dynamically and instantly providing a feedback and updating a recognition method for the face recognition.

Abstract: A model constructing method for audio recognition is provided. In the method, audio data is obtained. A predicted result of the audio data is determined by using the classification model which is trained by machine learning algorithm. The predicted result includes a label defined by the classification model. A prompt message is provided according to a loss level of the predicted result. The loss level is related to a difference between the predicted result and a corresponding actual result. The prompt message is used to query a correlation between the audio data and the label. The classification model is modified according to a confirmation response of the prompt message, and the confirmation response is related to a confirmation of the correlation between the audio data and the label. Accordingly, the labeling efficiency and predicting correctness can be improved.

Abstract: A processing method of face recognition includes steps of: extracting embedding feature information from a face image; outputting a recognition result of face recognition according to the embedding feature information, wherein the recognition result includes a recognized name and embedding feature distance information; determining whether the recognized name is in a list or not; if the recognized name is in the list, performing a removal checking step for determining whether to remove the recognition result based on the embedding feature distance information; if determining that the recognition result is not to be removed, displaying the recognized name; if determining that the recognition result is to be removed, displaying a negative prompt; and dynamically and instantly providing a feedback and updating a recognition method for the face recognition.

Abstract: A location and tracking system has wireless base stations, an input interface, and a processor. It receives initial physical environment inputs for an environment, with default signal attenuation values for wall and obstacle segments. An initial channel parameters fingerprint of projected signal strengths in the environment is generated according to the initial physical environment model. The processor simulates motion of the node along a hth trajectory to create a corresponding fitness function describing how similar the channel parameters estimated along the hth trajectory by the prediction model are to gathered channel parameters by the mobile node along its trajectory. The model and the fingerprint are subsequently modified in iterations as possible mobile node trajectories are determined to perform location error estimation which takes into account the re-calculated fingerprint and gathered channel parameters and the hth trajectory as a ground truth.

Abstract: A location and tracking system (1) comprises a site survey tool (2), a location engine (3), a self-calibration engine (4), and a fingerprint database (5). The system (1) operates for a wireless network of access points in a building and it generates from initialization data a model of the physical environment. The model may for example be in the form of data representing geometry of lines in plan and signal attenuation values for the lines. The system then uses this model to generate a fingerprint of projected signal strengths in the environment. The system may maintains a fingerprint for every environment model or there may be only one model at any one time, or there may be a number. Where the latter, the system may assign a weighting to each model to indicate its confidence value.