Autonomous Camera Platform for Aesthetic Image Pre-Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Users face challenges in capturing and selecting images that evoke strong emotional responses, as existing technologies limit the ability to take diverse and high-quality images across various times and locations, and require manual review of numerous images for social media sharing.
Innovation Solution
A system utilizing a movable platform equipped with a camera that autonomously captures images at different locations and orientations, calculating image scores based on aesthetics, saliency, and other aspects to automatically select and store the most appealing images, reducing the user's burden in image selection and sharing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If a photographer manually selects and takes pictures, then the quality and relevance of images are improved, but the quantity and diversity of images are limited
Solution Approach 1:
The system enables autonomous image capture by a movable platform that automatically navigates to optimal positions and captures images without continuous human intervention. The platform independently executes the photography task while the user defines only the target object and parameters.
Solution Approach 2:
The system pre-calculates optimal photographing positions and trajectories before actual image capture. The movable platform plans its movement path and capture points in advance, enabling efficient automated photography while maintaining quality standards.
2Measurement precision
If a user reviews all captured images manually, then the selection accuracy is improved, but the time and effort required increase significantly
Solution Approach 1:
The system employs computational models that learn from user feedback on previously selected images. When users rate selected images, the system adjusts its selection algorithm to better predict user preferences, improving accuracy over time while reducing manual review burden.
Solution Approach 2:
The system replaces manual image review with automated computational models including deep learning-based aesthetic evaluation. These algorithms objectively assess image quality and predicted social media performance, substituting human time-consuming review with rapid automated scoring.
3Productivity
If computational models predict image popularity, then the selection of shareable images is improved, but the perfect prediction accuracy is not achieved
Solution Approach 1:
The system uses multiple competing computational models with different parameter sets and architectural approaches to predict image popularity. By ensembling diverse models with varying parameters, the system achieves more robust predictions that compensate for individual model limitations.
Solution Approach 2:
The computational models continuously learn and adapt from actual social media performance data. The system automatically updates its prediction algorithms based on real-world feedback, improving prediction reliability over time without requiring manual recalibration.
4Quantity of substance
If multiple cameras are positioned at different locations to capture diverse images, then the quantity and variety of images are improved, but the system complexity and cost increase
Solution Approach 1:
The system replaces multiple static cameras with a single movable platform that dynamically changes its position and orientation. This dynamic approach achieves the same diversity capture capability as multiple fixed cameras but with reduced system complexity and cost.
Data Source
Figure 1
Figure 2
AI summary
The invention regards a method and respective system for assisting a user in producing and selecting images. When moving (S1) a moveable platform on which a camera is mounted, a plurality of images is captured with the camera. At least one image score for at least one region of the image is calculated in a processor (S5), and those images, for which the at least one image score fulfils a selection condition (S6), or identifiers of such images, are stored in a list in a memory. The stored images or the identifiers that allow retrieval of the stored images are output via an interface (S10).