18555136. METHOD AND SYSTEM FOR PROVIDING LABELED IMAGES FOR SMALL CELL SITE SELECTION simplified abstract (Telefonaktiebolaget LM Ericsson (publ))
Contents
METHOD AND SYSTEM FOR PROVIDING LABELED IMAGES FOR SMALL CELL SITE SELECTION
Organization Name
Telefonaktiebolaget LM Ericsson (publ)
Inventor(s)
Soufiane Belharbi of Montreal (CA)
Aydin Sarraf of Pierrefonds (CA)
METHOD AND SYSTEM FOR PROVIDING LABELED IMAGES FOR SMALL CELL SITE SELECTION - A simplified explanation of the abstract
This abstract first appeared for US patent application 18555136 titled 'METHOD AND SYSTEM FOR PROVIDING LABELED IMAGES FOR SMALL CELL SITE SELECTION
The disclosure pertains to a method for training a model to label images, involving obtaining a dataset of labeled images, selecting a second dataset of images for pseudo labeling, feeding these images into the model, combining the datasets, training the model, and testing it for performance.
- The method involves using a first dataset of labeled images selected from a pool of images showing different views of geographical areas.
- A second dataset of images is chosen from the pool for pseudo labeling, distinct from the first dataset.
- The second dataset is fed into the model to obtain pseudo-labels for the images.
- The first dataset and the second dataset with pseudo-labels are combined into a third dataset for training the model.
- The model is tested with a fourth dataset of labeled images, and if the desired performance is achieved, the model is stored; otherwise, the method is repeated.
Potential Applications: - This method can be applied in various industries such as autonomous vehicles, surveillance systems, and satellite imagery analysis. - It can enhance the accuracy and efficiency of image labeling tasks in computer vision applications.
Problems Solved: - Addresses the challenge of training models for image labeling with limited labeled data. - Improves the performance of models by incorporating pseudo-labeling techniques.
Benefits: - Increases the accuracy of image labeling models. - Enhances the generalization capabilities of models by leveraging pseudo-labeling. - Optimizes the training process by combining labeled and pseudo-labeled datasets.
Commercial Applications: - This technology can be utilized in industries such as agriculture, urban planning, and environmental monitoring for efficient image analysis and labeling tasks.
Prior Art: - Researchers have explored the use of pseudo-labeling techniques in machine learning to improve model performance with limited labeled data.
Frequently Updated Research: - Stay updated on advancements in pseudo-labeling techniques for image labeling tasks to enhance model performance and efficiency.
Questions about Image Labeling with Pseudo-Labeling: 1. How does pseudo-labeling improve the training process of image labeling models?
- Pseudo-labeling helps incorporate unlabeled data into the training process, enhancing model performance and generalization.
2. What are the potential challenges associated with using pseudo-labeling in image labeling tasks?
- Some challenges include ensuring the quality of pseudo-labels and managing the integration of labeled and pseudo-labeled datasets effectively.
Original Abstract Submitted
The disclosure relates to a method for training a model for labeling images. The method comprises obtaining a first dataset of labeled images, the first dataset of labeled images being selected from a pool of images presenting different views of portions of geographical areas; selecting, from the pool of images, a second dataset of images for pseudo labeling, different from the first dataset; feeding the second dataset of images into the model and obtaining as output of the model the second dataset of images with pseudo-labels; combining the first dataset of labeled images with the second dataset of images with pseudo-labels into a third dataset; training the model with the third dataset; and testing the model with a fourth dataset of labeled images and, upon determining that a requested performance is met, storing the model, or, upon determining that the requested performance is not met, executing the method again.
