Acne Type Localization And Counting CNN For Small Facial Lesions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies struggle to accurately localize, count, and visualize acne on facial images, particularly for different types, which is crucial for dermatological applications and e-commerce recommendations.
Innovation Solution
A convolutional neural network (CNN) model is used to process facial images end-to-end, identifying and localizing acne types without cropping, and generating anchor boxes for precise localization, with k-means clustering for anchor box aspect ratio calculation, and non-maximum suppression for redundant detection filtering.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a CNN model is used to detect small acne objects in facial images, then detection precision is improved, but the complexity of the detection system increases
Solution Approach 1:
The detection system is segmented into multiple specialized components: a skin region segmentation module that isolates skin areas from the facial image, and an acne detection module that focuses specifically on detecting acne within the segmented skin regions. This segmentation allows the system to achieve high detection precision for small acne objects without requiring the entire system to be overly complex, as each module has a dedicated function.
Solution Approach 2:
The patent introduces an intermediary skin mask as a mediator between the input facial image and the acne detection process. The skin mask serves as an intermediate representation that highlights only the skin regions, making it easier for the detection model to identify acne without being distracted by non-skin elements. This intermediary structure simplifies the detection task while maintaining high precision.
2Loss of information
If the model processes the entire facial image without cropping, then no information is lost, but the computational cost increases
Solution Approach 1:
The patent extracts only the relevant skin regions from the entire facial image using a skin segmentation module. By taking out and isolating the skin areas where acne can occur, the system avoids processing unnecessary non-skin regions (such as hair, eyes, mouth). This extraction reduces the computational burden while ensuring that all potential acne locations are captured, thus minimizing information loss.
3Measurement precision
If multiple acne types are detected and classified, then diagnostic accuracy is improved, but the complexity of classification increases
Solution Approach 1:
The classification system applies local quality by using different classification strategies for different acne types. The model is trained to recognize specific visual characteristics of each acne type (such as inflammatory acne with redness and swelling, comedonal acne with visible pores, pigmentary acne with dark spots) and applies appropriate classification criteria locally to each detected acne instance. This allows high diagnostic accuracy without requiring a uniformly complex classification structure throughout the entire system.
Data Source
AI summary
Systems, methods and techniques provide for acne localization, counting and visualization. An image is processed using a trained model to identify objects. The model may be a deep learning (e.g. convolutional neural) network configured for object classification with a detection focus on small objects. The image may be a frontal or profile facial image, processed end to end. The model identifies and localizes different types of acne. Instances are counted and visualized such as by annotating the source image. An example annotation is an overlay identifying a type and location of each instance. Counts by acne type assist with scoring. A product and/or service may be recommended in response to the identification of the acne (e.g. the type, localization, counting and/or a score).