公开(公告)号：US20160366323A1

公开(公告)日：2016-12-15

申请号：US15088070

申请日：2016-03-31

Applicant: Mediatek Inc.

Inventor： Cheng-Che CHAN ,  Cheng-Che CHEN ,  Po-hua HUANG

IPC: H04N5/235 ,  H04N5/247 ,  H04N5/265 ,  G06T1/20 ,  H04N5/232 ,  G06T15/50

CPC classification number: G06T15/50 ,  G06T2200/04 ,  H04N5/2258 ,  H04N5/23216 ,  H04N5/235

Abstract: Methods and systems for improving images and video captured by a device is provided. The methods and systems may involve a virtual 3D model for a scene that is in view of a device's sensors. The virtual 3D model may be generated using the sensor information and be used to produce a 2D lighting image. The lighting image may be applied to a captured image of the scene to improve lighting of the captured image or video. A virtual light source may be implemented as part of the process, which if desired, can be moved, adjusted, or modified in the virtual 3D model to adjust the lighting image and consequently adjust alighted image or a final image.

Abstract translation: 提供了用于改善由设备捕获的图像和视频的方法和系统。 所述方法和系统可以涉及针对设备传感器的场景的虚拟3D模型。 虚拟3D模型可以使用传感器信息生成并用于产生2D照明图像。 照明图像可以应用于场景的拍摄图像，以改善拍摄图像或视频的照明。 虚拟光源可以被实现为处理的一部分，如果需要，可以在虚拟3D模型中移动，调整或修改虚拟光源，以调整照明图像，从而调整下载图像或最终图像。

公开(公告)号：US20240361999A1

公开(公告)日：2024-10-31

申请号：US18309341

申请日：2023-04-28

Applicant: MEDIATEK INC.

Inventor： Meng-Hsuan YANG ,  Po-hua HUANG ,  Hsing-Chang CHOU ,  Ting Chen TSAN ,  Yu-Lung LU

IPC: G06F8/41 ,  G06N3/10

CPC classification number: G06F8/443 ,  G06N3/10

Abstract: A method of compiling a deep learning model includes reading metadata from a compiled result, the metadata indicating a structure of the deep learning model corresponding to a low-level IR, receiving shape information of an input tensor of the deep learning model, determining a shape of an output tensor of a first computation operation of the computation operations based on the shape information of the input tensor of the deep learning model and the structure of the deep learning model, tiling the output tensor of the first computation operation into one or more tiles according to the shape of the output tensor of the first computation operation and hardware limitations of a processor executing the deep learning model, and patching one or more copies of a templated hardware command into executable hardware commands.