Abstract: A capacitive ultrasonic transducer device includes a substrate, a first capacitive structure, a second capacitive structure, a first film structure and a second film structure. The first capacitive structure is disposed on the substrate, and includes a first electrode and a second electrode. A first gap and a dielectric layer are located between the first electrode and the second electrode. The second capacitive structure is disposed on the substrate, and includes a third electrode and a fourth electrode. A second gap is located between the third electrode and the fourth electrode. The first film structure is configured to seal the first gap. The second film structure is connected to the third electrode and the fourth electrode, and configured to seal the second gap. A first width between the first electrode and the second electrode is different from a second width of the second gap.

Abstract: A chip packaging structure and a method for fabricating the same are provided. The chip package structure includes a conductive substrate, a dam and a metal shielding layer. The conductive substrate includes a substrate, vias and electrodes. The substrate has first and second board surfaces opposite to each other. The vias penetrate through the first board surface and the second board surface, and a part of the vias is disposed in a first die-bonding region on which a chip is to be arranged. The electrodes extend from the first board surface to the second board surface through the vias. The dam is formed on the first board surface to surround the first die-bonding region, and the dam has a height higher than that of the chip. The metal shielding layer covers the dam and a part of the first board surface that do not overlap with the electrodes.

Abstract: An electronic device and a method for training a neural network model are provided. The method includes: obtaining a first neural network model and a first pseudo-labeled data; inputting the first pseudo-labeled data into the first neural network model to obtain a second pseudo-labeled data; determining whether a second pseudo-label corresponding to the second pseudo-labeled data matching a first pseudo-label corresponding to the first pseudo-labeled data; in response to the second pseudo-label matching the first pseudo-label, adding the second pseudo-labeled data to a pseudo-labeled dataset; and training the first neural network model according to the pseudo-labeled dataset.

Abstract: An anomaly detection device based on a generative adversarial network architecture, which uses the single-type training data composed of multiple normal signals to train an anomaly detection model. The anomaly detection device includes an encoder, a generator, a discriminator, and a random vector generator. In the training phase of anomaly detection model, the random latent vectors generated by the random vector generator are sequentially input to a generator to generate the synthesized signals with the same dimension as the normal signals. The synthesized signals are sequentially input into a discriminator to output the corresponding discriminant values. When the corresponding discriminant values are under the predetermined threshold, the corresponding synthesized signals are selected as the anomalous class training data, and the real normal signals are selected as the normal class training data.

Abstract: An interactive object retrieval method is provided. The present method includes receiving a time-space searching condition and a query, and selecting a plurality of searching results from an object database in accordance with the time-space searching condition, a similarity between the query and each of a plurality of data records of a first category in the object database, and a time information and a location information corresponding to each of a plurality of data records of a second category in the object database. The method further includes receiving at least one user input corresponding to at least one of the searching results, and determining a display manner of the searching results on a user interface in accordance with the at least one user input and the similarity between the query and each searching result.

Abstract: An automatic image classification method applying an unsupervised adaptation method is provided. A plurality of non-manually-labeled observation data are grouped into a plurality of groups. A respectively hypothesis label is set to each of the groups according to a classifier. It is determined whether each member of the observation data in each of the groups is suitable for adjusting the classifier according to the hypothesis label, and the non-manually-labeled observation data which are determined as being suitable for adjusting the classifier are set as a plurality of adaptation data. The classifier is updated according to the hypothesis label and the adaptation data. The observation data are classified according to the updated classifier.

Abstract: In one exemplary embodiment, an object region tracking and picturing module is constructed on a moving platform of a mobile end and a remote control module is constructed on another platform for an image object region tracking system. The two modules communicate with each other via a digital network for delivering required information. The object region tracking and picturing module uses a real-time image backward search technology to store at least an image frame previously captured on the moving platform into a frame buffer, and start tracking an object region from the position pointed out by the remote control module to a newest image frame captured on the moving platform, then find out a relative position on the newest image frame for the tracked object region.

Abstract: An image detecting method and a system thereof are provided. The image detecting method includes the following steps. An original image is captured. A moving-object image of the original image is created. An edge-straight-line image of the original image is created, wherein the edge-straight-line image comprises a plurality of edge-straight-lines. Whether the original image has a mechanical moving-object image is detected according to the length, the parallelism and the gap of the part of the edge-straight-lines corresponding to the moving-object image.

Abstract: An interactive object retrieval method is provided. The present method includes receiving a time-space searching condition and a query, and selecting a plurality of searching results from an object database in accordance with the time-space searching condition, a similarity between the query and each of a plurality of data records of a first category in the object database, and a time information and a location information corresponding to each of a plurality of data records of a second category in the object database. The method further includes receiving at least one user input corresponding to at least one of the searching results, and determining a display manner of the searching results on a user interface in accordance with the at least one user input and the similarity between the query and each searching result.

Abstract: An automatic image classification method applying an unsupervised adaptation method is provided. A plurality of non-manually-labeled observation data are grouped into a plurality of groups. A respectively hypothesis label is set to each of the groups according to a classifier. It is determined whether each member of the observation data in each of the groups is suitable for adjusting the classifier according to the hypothesis label, and the non-manually-labeled observation data which are determined as being suitable for adjusting the classifier are set as a plurality of adaptation data. The classifier is updated according to the hypothesis label and the adaptation data. The observation data are classified according to the updated classifier.

Abstract: In one exemplary embodiment, a tracking system for region-of-interest (ROI) performs a feature-point detection locally on an ROI of an image frame at an initial time via a feature point detecting and tracking module, and tracks the detected features. A linear transformation module finds out a transform relationship between two ROIs of two consecutive image frames, by using a plurality of corresponding feature points. An estimation and update module predicts and corrects a moving location for the ROI at a current time. Based on the result corrected by the estimation and update module, an outlier rejection module removes at least an outlier outside the ROI.

Abstract: A semiconductor chip assembly includes a semiconductor device, a heat spreader, a conductive trace, an adhesive and a support layer. The heat spreader includes a post, a base, an underlayer and a thermal via. The conductive trace includes a pad and a terminal. The semiconductor device is electrically connected to the conductive trace and thermally connected to the heat spreader. The post extends upwardly from the base into an opening in the adhesive, the base extends laterally from the post, the support layer is sandwiched between the base and the underlayer and the thermal via extends from the base through the support layer to the underlayer. The conductive trace provides signal routing between the pad and the terminal.

Abstract: An image processing method and a system are provided. The image processing method of moving camera comprises the following steps. An image of a road is captured by a first camera unit. A coordinate of the image of an object shown in the image of the road is captured when the image of the object shown in the image of the road is selected. At least an aiming angle of a second camera unit is adjusted according to the coordinate to make the field-of-view of the second camera unit aligned with the object. The image of the object is captured by the second camera unit. The image of the object is enlarged.

Abstract: An apparatus and method for moving object detection computes a corresponding frame difference for every two successive image frames of a moving object, and segments a current image frame of the two successive image frames into a plurality of homogeneous regions. At least a candidate region is further detected from the plurality of homogeneous regions. The system gradually merges the computed frame differences via a morphing-based technology and fuses the merged frame difference with the at least a candidate region, thereby obtaining the location and a complete outline of the moving object.

Abstract: A pixel region-based image segmentation method is disclosed. When an input image is retrieved, pixels thereof are sequentially scanned row by row. Signs of unmarked pixels of the input image are determined according to region features of neighboring pixels of each pixel and pixel update information is recorded to generate a region sign update data table and a region sign feature data table. The pixels of the input frame are further scanned row by row to retrieve signs of the pixels and region signs of the pixels are determined and updated according to the region sign update data table.

Abstract: A method for tracking a moving object is provided. The method detects the moving object in a plurality of continuous images so as to obtain space information of the moving object in each of the images. In addition, appearance features of the moving object in each of the images are captured to build an appearance model. Finally, the space information and the appearance model are combined to track a moving path of the moving object in the images. Accordingly, the present invention is able to keep tracking the moving object even if the moving object leaves the monitoring frame and returns again, so as to assist the supervisor in finding abnormal acts and making following reactions.

Abstract: In one exemplary embodiment, an object region tracking and picturing module is constructed on a moving platform of a mobile end and a remote control module is constructed on anther platform for an image object region tracking system. The two modules communicate with each other via a digital network for delivering required information. The object region tracking and picturing module uses a real-time image backward search technology to store at least an image frame previously captured on the moving platform into a frame buffer, and start tracking an object region from the position pointed out by the remote control module to a newest image frame captured on the moving platform, then find out a relative position on the newest image frame for the tracked object region.

Abstract: Disclosed is a moving object detection apparatus and method by using optical flow analysis. The apparatus includes four modules of image capturing, image aligning, pixel matching, and moving object detection. Plural images are successively inputted under a camera. Based on neighboring images, frame relationship on the neighboring images is estimated. With the frame relationship, a set of warping parameter is further estimated. Based on the wrapping parameter, the background areas of the neighboring images are aligned to obtain an aligned previous image. After the alignment, a corresponding motion vector for each pixel on the neighboring images is traced. The location in the scene of the moving object can be correctly determined by analyzing all the information generated from the optical flow.

Abstract: A method of making a semiconductor chip assembly includes providing a post, a base, a support layer and an underlayer, wherein the post extends above the base and the support layer is sandwiched between the base and the underlayer, mounting an adhesive on the base including inserting the post into an opening in the adhesive, mounting a conductive layer on the adhesive including aligning the post with an aperture in the conductive layer, then flowing the adhesive upward between the post and the conductive layer, solidifying the adhesive, then providing a conductive trace that includes a pad, a terminal and a selected portion of the conductive layer, providing a heat spreader that includes the post, the base, the underlayer and a thermal via that extends from the base through the support layer to the underlayer, then mounting a semiconductor device on the post, electrically connecting the semiconductor device to the conductive trace and thermally connecting the semiconductor device to the heat spreader.

Abstract: In one exemplary embodiment, a tracking system for region-of-interest (ROI) performs a feature-point detection locally on an ROI of an image frame at an initial time via a feature point detecting and tracking module, and tracks the detected features. A linear transformation module finds out a transform relationship between two ROIs of two consecutive image frames, by using a plurality of corresponding feature points. An estimation and update module predicts and corrects a moving location for the ROI at a current time. Based on the result corrected by the estimation and update module, an outlier rejection module removes at least an outlier outside the ROI.