Lens Tracking Offset via Deep Learning Composition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automatic target tracking in photography, particularly in portrait capturing, is limited by the 'center control' method using bounding boxes, which fails to adapt to different target postures and compositions, resulting in suboptimal capturing effects.
Innovation Solution
A method and apparatus for image capturing that utilize a pre-trained reference model based on deep convolutional neural networks to predict the optimal composition position of a target within an image, calculating pixel-level movement offsets to adjust the camera, thereby improving capturing effects across various postures and angles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If the 'center control' method using bounding box is used for automatic target tracking, then the implementation is simple and can be applied to various scenes, but the capturing effect is suboptimal for portrait photography and does not adapt to different target postures
Solution Approach 1:
The patent replaces the traditional mechanical bounding box control method with a deep learning-based pixel-level feature recognition system. The system uses convolutional neural networks to analyze image content and determine optimal composition positions, substituting simple geometric control with intelligent visual analysis that adapts to different postures and scenes.
Solution Approach 2:
The patent changes the control parameters from simple bounding box coordinates to pixel-level visual features and composition scores. By analyzing features such as subject position, posture, and composition quality at the pixel level, the system dynamically adjusts tracking parameters to achieve optimal capturing effects for different scenarios.
2Manufacturing precision
If the 'center control' method is used, then the system is easy to operate, but the composition quality and capturing effect are limited
Solution Approach 1:
The system performs self-service by automatically analyzing image content and adjusting composition without user intervention. The deep learning model autonomously evaluates different composition options and selects the optimal one, eliminating the need for manual camera control while achieving high composition precision.
Solution Approach 2:
The patent implements a feedback mechanism where the system continuously analyzes the captured image, evaluates composition quality based on pixel-level features, and adjusts the camera position accordingly. This closed-loop control ensures high composition precision while maintaining ease of operation through automatic adjustment.
Data Source
AI summary
Provided are a method, apparatus and device for image capturing and a storage medium. The method includes acquiring the bounding box of a lens tracking target in an image to be captured; using a pre-trained reference model to predict the first reference position of the image to be captured; and determining a lens movement offset based on the position of each pixel in the bounding box and the first reference position.