Abstract: A method of performing automatic tuning on a deep learning model includes: utilizing an instruction-based learned cost model to estimate a first type of operational performance metrics based on a tuned configuration of layer fusion and tensor tiling; utilizing statistical data gathered during a compilation process of the deep learning model to determine a second type of operational performance metrics based on the tuned configuration of layer fusion and tensor tiling; performing an auto-tuning process to obtain a plurality of optimal configurations based on the first type of operational performance metrics and the second type of operational performance metrics; and configure the deep learning model according to one of the plurality of optimal configurations.

Abstract: A circuit board structure includes a circuit substrate having opposing first and second sides, a redistribution structure disposed at the first side, and a dielectric structure disposed at the second side. The circuit substrate includes a first circuit layer disposed at the first side and a second circuit layer disposed at the second side. The redistribution structure is electrically coupled to the circuit substrate and includes a first leveling dielectric layer covering the first circuit layer, a first thin-film dielectric layer disposed on the first leveling dielectric layer and having a material different from the first leveling dielectric layer, and a first redistributive layer disposed on the first thin-film dielectric layer and penetrating through the first thin-film dielectric layer and the first leveling dielectric layer to be in contact with the first circuit layer. The dielectric structure includes a second leveling dielectric layer disposed below the second circuit layer.

Abstract: A multi-camera electronic device and a control method thereof are proposed. The method includes the following steps. At least one camera of the electronic device is used for scene detection to generate photographing analysis information. All the photographing analysis information is collected, and joint photographing information including a joint target is generated through a communication process among all the cameras. An individual photographing parameter of each camera is generated according to the joint photographing information. Each camera is controlled to capture an image of the scene by using its individual photographing parameter to respectively generate a corresponding output image.

Abstract: An electronic apparatus and a method of generating a depth map are provided. The electronic apparatus includes a projection device, a first camera, a second camera, and a processing device. The projection device projects patterned invisible light on a scene. The first camera captures visible light information to generate a first image. The second camera captures visible light information and invisible light information to generate a second image. The processing device receives the first image and the second image. A first depth information unit in the processing device compares visible light images of the first image and the second image to obtain first depth information, and a second depth information unit in the processing device identifies an invisible light image in the second image to obtain second depth information. The processing device generates a depth map of the scene from the first depth information and the second depth information selectively.

Abstract: A multi-camera electronic device and a control method thereof are proposed. The method includes the following steps. At least one camera of the electronic device is used for scene detection to generate photographing analysis information. All the photographing analysis information is collected, and joint photographing information including a joint target is generated through a communication process among all the cameras. An individual photographing parameter of each camera is generated according to the joint photographing information. Each camera is controlled to capture an image of the scene by using its individual photographing parameter to respectively generate a corresponding output image.

Abstract: An electronic apparatus and a method of generating a depth map are provided. The electronic apparatus includes a projection device, a first camera, a second camera, and a processing device. The projection device projects patterned invisible light on a scene. The first camera captures visible light information to generate a first image. The second camera captures visible light information and invisible light information to generate a second image. The processing device receives the first image and the second image. A first depth information unit in the processing device compares visible light images of the first image and the second image to obtain first depth information, and a second depth information unit in the processing device identifies an invisible light image in the second image to obtain second depth information. The processing device generates a depth map of the scene from the first depth information and the second depth information selectively.

Abstract: The present disclosure illustrates a camera array correction method. The method comprises steps of calculating the gain values between the images captured by cameras with the pre-measured characteristic of each of the cameras. By such correction method, the different cameras in the camera array can capture images of the same brightness values and color values simultaneously. In addition, the huge computation requirement due to the increasing of the cameras can be reduced effectively.