Gaussian Process Assisted Meta-learning for Image Classification and Object Detection Models
This addresses the challenge of efficiently collecting data for computer vision models, particularly in scenarios with rare objects or varied conditions, offering a practical tool for domain-specific applications.
The paper tackles the problem of costly data collection for machine learning models by proposing a meta-learning approach that uses Gaussian process surrogates to guide data acquisition based on metadata, resulting in improved model performance compared to random selection, as demonstrated on image classification and object detection tasks including an aerial image application.
Collecting operationally realistic data to inform machine learning models can be costly. Before collecting new data, it is helpful to understand where a model is deficient. For example, object detectors trained on images of rare objects may not be good at identification in poorly represented conditions. We offer a way of informing subsequent data acquisition to maximize model performance by leveraging the toolkit of computer experiments and metadata describing the circumstances under which the training data was collected (e.g., season, time of day, location). We do this by evaluating the learner as the training data is varied according to its metadata. A Gaussian process (GP) surrogate fit to that response surface can inform new data acquisitions. This meta-learning approach offers improvements to learner performance as compared to data with randomly selected metadata, which we illustrate on both classic learning examples, and on a motivating application involving the collection of aerial images in search of airplanes.