Visibility-Adaptive ANN Resolution for Edge Image Analysis

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

ANN-based image analysis solutions require significant computational resources, making them unsuitable for edge devices with limited resources, especially when multiple tasks need to be performed under varying visibility conditions like fog and smog.

Innovation Solution

Adaptive selection of ANN architectures based on visibility conditions, using lower-resolution architectures for reduced visibility scenarios to conserve computational resources, and optionally downsampling images to match the selected architecture.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a higher-resolution ANN architecture is used for image analysis, then analysis accuracy is improved, but computational resource consumption increases

Engineering Contradiction:

Improveanalysis accuracyVSAvoidcomputational resource consumption

Core Design Contradiction:

Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent implements dynamic selection of ANN architecture resolution based on real-time visibility conditions. The system adapts between high-resolution architectures (for clear visibility) and low-resolution architectures (for foggy/smoggy conditions), making the computational resource allocation dynamic rather than static. This resolves the contradiction by matching the level of computational complexity to the actual information content available in the input image.

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The patent changes the resolution parameter of the ANN architecture based on visibility conditions. When visibility is poor (foggy/smoggy), the system selects ANN architectures with lower input resolution parameters, reducing computational complexity while maintaining adequate analysis accuracy for the given conditions. This parameter adaptation directly addresses the trade-off between accuracy and resource consumption.

Inventive Principle:

Principle #35Parameter changes

2Adaptability or versatility

If multiple ANN architectures are maintained for different visibility conditions, then adaptability to varying conditions is improved, but device complexity increases

Engineering Contradiction:

Improveadaptability to visibility conditionsVSAvoidsystem complexity

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent segments the ANN architecture selection into discrete categories (high-resolution for clear visibility, low-resolution for poor visibility). This segmentation allows the system to maintain a manageable set of pre-defined architectures rather than implementing a continuous adaptation mechanism, reducing overall system complexity while preserving adaptability to different visibility conditions.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces a visibility assessment module as an intermediary that evaluates current visibility conditions and selects the appropriate ANN architecture. This intermediary layer simplifies the complexity by centralizing the decision-making logic and isolating the complexity of multiple architectures behind a simple selection interface, making the overall system more manageable.

Inventive Principle:

Principle #24Intermediary (Mediator)

3Productivity

If image resolution is reduced to match lower-resolution ANN architecture, then computational processing speed is improved, but information detail is lost

Engineering Contradiction:

Improveprocessing speedVSAvoidimage detail information

Core Design Contradiction:

ProductivityVSLoss of information

Solution Approach 1:

The patent applies partial action by using reduced-resolution image processing only when visibility conditions warrant it (foggy/smoggy conditions). In clear visibility conditions, the full-resolution processing is maintained. This partial application of resolution reduction minimizes information loss while still achieving processing speed improvements when actually needed.

Inventive Principle:

Principle #16Partial or excessive action

Solution Approach 2:

The patent converts the harmful effect of fog and smog (which naturally reduce visible detail in the scene) into a benefit by justifying reduced image resolution processing. Since the scene already contains less visible information due to atmospheric conditions, reducing image resolution for processing aligns with the actual information available, preventing unnecessary processing of obscured details while maintaining processing speed.

Inventive Principle:

Principle #22Blessing in disguise (Convert harm into benefit)

Data Source

PatentUS20260004580A1Neural network-based analysis of images captured under different visibility conditions

Publication Date: 2026.01.01 AXIS

AI summary

A solution for analyzing images of a scene captured under different visibility conditions includes obtaining images of a scene captured by one or more cameras and, for each image, obtaining an indication of an actual or assumed visibility (distance) at the scene when the image was captured; selecting, based on the visibility, an artificial neural network (ANN) architecture from a plurality of ANN architectures, wherein the ANN architectures are each trained for image analysis but configured for different input image resolutions, and analyzing the image using the selected ANN architecture. If the selected ANN architecture has a lower input image resolution than the image, the image may be downscaled to match the input image resolution of the selected ANN architecture.