Systems, apparatuses, and methods for rapid machine learning for floor segmentation for robotic devices are disclosed herein. According to at least one non-limiting exemplary embodiment, a robotic system is disclosed. The robotic system may comprise a neural network embodied therein capable of learning associations between color values of pixels and corresponding classifications of those pixels, wherein neural network is trained initially to identify floor and non-floor pixels within images. A user input may be provided to the neural network to further configure the neural network to be able to identify navigable floors and unnavigable floors unique to an environment without a need for additional annotated training images specific to the environment.

A diagnostic tool for deep learning similarity models and image classifiers provides valuable insight into neural network decision-making. A disclosed solution generates a saliency map by: receiving a baseline image and a test image; determining, with a convolutional neural network (CNN), a first similarity between the baseline image and the test image; based on at least determining the first similarity, determining, for the test image, a first activation map for at least one CNN layer; based on at least determining the first similarity, determining, for the test image, a first gradient map for the at least one CNN layer; and generating a first saliency map as an element-wise function of the first activation map and the first gradient map. Some examples further determine a region of interest (ROI) in the first saliency map, cropping the test image to an area corresponding to the ROI, and determine a refined similarity score.

A neural network training method, an image processing method, and apparatuses thereof are provided. The neural network training method includes obtaining a first domain image and a second domain image, where the first domain image and the second domain image are unpaired images in different domains; obtaining a scaled first domain image by scaling, at an iteration, the first domain image; obtaining a training patch by cropping the scaled first domain image, where each training patch has a same number of pixels with different contents; inputting the training patch into the neural network at the iteration, and outputting an output patch; calculating a contrastive loss based on a query sub-patch and negative sub-patches selected from the training patch and a corresponding positive sub-patch selected from the output patch; and updating model parameters of the neural network based on the contrastive loss and a generative adversarial network loss.

Methods and systems for data collection in an environment including pumps and fans are disclosed. A monitoring system may include a data collector communicatively coupled to a plurality of input channels, wherein the input channels are communicatively coupled to sensors measuring operational parameters of a pump or fan. A data storage may store one or more frequencies related to an operation of the pump or fan, and a data acquisition circuit may interpret a plurality of detection values from the collected data. A frequency evaluation circuit may detect a signal on one of the input channels at a frequency higher than the one or more frequencies at which the pump or fan operates.

An image processing apparatus acquires first image data obtained by capturing an image of a subject, second image data obtained by capturing an image of a person capturing the subject and surroundings thereof, and third image data indicating an appearance of an image capture apparatus that captures the image of the subject. The apparatus reduces a reflection in the first image data using a learned machine learning model that uses the first image data, the second image data, and the third image data as input data.

A system and method are provided for machine learning (ML) quality assurance. The method trains a plurality of agent ML annotation model software applications. Each agent annotation model is trained with a corresponding subset of annotated raw data images including annotation marks forming a boundary surrounding the first shape. A baseline ML annotation model is trained with all the subsets of annotated raw data images. The method accepts an evaluation dataset with unannotated images including the first shape, which is provided to the agent models and baseline models. In response to the evaluation dataset, the agent and baseline models infer predicted images including annotation marks forming a boundary surrounding the first shape. The baseline model predicted images are compared to the predicted images of each agent model for the purpose of determining agent model quality and identifying problematic raw data images for retraining purposes.

Embodiments of the present disclosure relate to a method, an electronic device, a storage medium and a program product for sample analysis. The method comprises: obtaining a sample set, the sample set being associated with annotation data; processing the sample set with a target model to determine prediction data for the sample set and confidence of the prediction data; determining accuracy of the target model based on a comparison between the prediction data and the annotation data; and determining a candidate sample which is potentially inaccurately annotated from the sample set based on the accuracy and the confidence. In this way, a potential inaccurately annotated sample may be efficiently screened out.

A system includes a memory; and a processor configured to train a first machine learning model based on the first dataset labeling; provide the second dataset to the trained first machine learning model to generate an updated second dataset including an updated second dataset labeling, determine a first difference between the updated second dataset labeling and the second dataset labeling; train a second machine learning model based on the updated second dataset labeling if the first difference is greater than a first threshold value; provide the first dataset to the trained second machine learning model to generate an updated first dataset including an updated first dataset labeling, determine a second difference between the updated first dataset labeling and the first dataset labeling; and train the first machine learning model based on the updated first dataset labeling if the second difference is greater than a second threshold value.

A method for training a deep learning network for face recognition includes: utilizing a face landmark detector to perform face alignment processing on at least one captured image, thereby outputting at least one aligned image; inputting the at least one aligned image to a teacher model to obtain a first output vector; inputting the at least one captured image a student model corresponding to the teacher module to obtain a second output vector; and adjusting parameter settings of the student model according to the first output vector and the second output vector.

At least one processor of an apparatus functions as a generation unit that identifies at least an outer edge of a specific region in a surface layer of an object and that generates outer edge candidates, and a control unit that selects an outer edge candidate based on an instruction from a user among the generated outer edge candidates.