Machine Learning Model Brightness Channel Modification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Machine learning models for image analysis perform poorly when analyzing user-submitted images with variable lighting conditions, as they rely heavily on brightness information that differs from the training data.
Innovation Solution
The system converts training data into a color space with a brightness channel, modifies the brightness channel values, and then reconverts it back to the original color space to retrain the model, de-emphasizing brightness information and enabling the model to rely on other image features for object identification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If machine learning models are trained using standard training data, then the model achieves good accuracy on training data, but the model performs poorly on user-submitted images with variable lighting conditions
Solution Approach 1:
The patent applies parameter changes by modifying the brightness channel values of training images. The system converts training images to a color space with a brightness channel, modifies the brightness values (e.g., adjusting illumination conditions), and then reconverts them to the original color space. This transforms the training data parameters to better match the variable lighting conditions of user-submitted images, enabling the model to generalize better across different lighting environments while maintaining training accuracy.
2Speed
If the model relies heavily on brightness information for object identification, then the model achieves fast and simple processing, but the model becomes sensitive to lighting condition variations
Solution Approach 1:
The patent implements feedback by using the modified brightness channel values during training to adjust the model's learned features. The system trains the machine learning model using the training images with modified brightness values, allowing the model to learn from the feedback of these adjusted parameters. This feedback mechanism enables the model to reduce its dependence on absolute brightness values while maintaining processing efficiency, as the model adapts its feature extraction based on the modified brightness information during training.
Data Source
AI summary
The present disclosure provides devices, systems and computer-readable media for identifying object characteristics using a machine learning model trained to de-emphasize brightness values. The machine learning model can be trained using modified training data. Modifying the training data can include converting the training data from an original color space into a color space having a brightness channel. The values of the brightness channel for the, training data can then be modified. After the values of the brightness channel are modified, the training data can be converted back into the original color space and used to train the machine learning model. A detection device can be configured with the machine learning model and used to identify object characteristics.