Examples of the present invention provides a method and system for inspection of heat recovery steam generator (HRSG) equipment to identify defects and damages using computer vision and deep learning techniques. The method comprising capturing one or more input frames by one or more input devices, classifying the one or more input frames by a scenario classifier to identify a scenario type based on a first modelled data prepared by training one or more deep neural networks (DNN), selecting at least one damage detector based on the identified scenario type, identifying one or more damage types by the at least one damage detector based on second modelled data prepared by training the one or more DNN and displaying one or more output frame indicating the identified one or more damage types of the HRSG equipment.

Techniques are described that facilitate training an artificial intelligence (AI) model in a manner that ensures equitable model performance across different sub-groups. According to an embodiment, a system is provided that comprises a memory that stores computer executable components, and a processor that executes the computer executable components stored in the memory. The computer executable components comprise a training component that trains a machine learning (ML) model on training data to perform an inferencing task using an equitable loss function that drives equitable performance of the ML model across different sub-groups represented the training data, resulting in trained version of the ML model that provides a defined equitable performance level across the different sub-groups. The equitable loss function is “sub-group aware” and penalizes variation in model performance across the sub-groups during model training and validation.

A method according to the present disclosure includes (b1) selecting a reference class from a plurality of classes, (b2) calculating a plurality of degrees of similarity between a feature spectrum corresponding to target training data and a plurality of the feature spectra belonging to the reference class, (b3) applying, to the plurality of degrees of similarity, a defectiveness function that is determined in advance, and calculating a defectiveness index with respect to the target training data, and (b4) determining whether the target training data is the defective data, based on a result of comparison between the defectiveness index and a threshold value.

An image processing device for generating learning data that is used for machine learning includes a processor that obtains image data. The processor specifies an unprocessable region that is a region in which a predetermined process cannot be performed or a region in which the predetermined process is not performed in an image region of the image data, and generates image data on which the predetermined process is performed in a region except the unprocessable region in the image region, as the learning data.

A method for multi-label image classification in a convolutional neural network (CNN) is provided that includes forming a composite image from a plurality of clipped images, and processing the composite image by the CNN to generate a probability vector for each clipped image of the plurality of clipped images, wherein a length of a probability vector is equal to a number of classes the CNN is designed to classify.

The present disclosures provides systems and methods for generating composite based data for use in machine learning systems, such as for use in training a machine learning system on the composite based data to identify an object of interest. In an aspect, a method of generating composite based data for use in training machine learning systems comprises: receiving a plurality of images, each of the plurality of images having a corresponding label; generating a composite image comprising the plurality of images, each of the plurality of images occupying a region of the composite image; generating a response map for the composite image, the response map having a plurality of response entries, each response entry encoded with a desired label corresponding to a fragment of the composite image, and generating composite data comprising the desired label of a response entry and image data corresponding to the fragment of the composite image.

System and methods for content- and contention-aware object detection are provided. A system may receive video information and perform object detection and object tracking based on an execution configuration. The system may approximate an optimized execution configuration. To approximate the optimized execution configuration, the system may identify, based on the video information, a plurality of content features. The system may further measure a contention level of a computer resource or multiple resources. The system may approximate, based on the content features and the utilization metric, latency metrics, for a plurality of execution configuration sets, respectively. The system may also approximate, based on the content features, accuracy metrics for the execution configuration sets, respectively. The system may select the optimized execution configuration set in response to satisfaction of a performance criterion. The system may perform object detection and object tracking based on the optimized execution configuration set.

The disclosure herein describes preparing and using a cross-attention model for action recognition using pre-trained encoders and novel class fine-tuning. Training video data is transformed into augmented training video segments, which are used to train an appearance encoder and an action encoder. The appearance encoder is trained to encode video segments based on spatial semantics and the action encoder is trained to encode video segments based on spatio-temporal semantics. A set of hard-mined training episodes are generated using the trained encoders. The cross-attention module is then trained for action-appearance aligned classification using the hard-mined training episodes. Then, support video segments are obtained, wherein each support video segment is associated with video classes. The cross-attention module is fine-tuned using the obtained support video segments and the associated video classes. A query video segment is obtained and classified as a video class using the fine-tuned cross-attention module.

Systems and methods for data fusion and analysis of vehicle sensor data, including receiving a multiple modality input data stream from a plurality of different types of vehicle sensors, determining latent features by extracting modality-specific features from the input data stream, and aligning a distribution of the latent features of different modalities by feature-level data fusion. Classification probabilities can be determined for the latent features using a fused modality scene classifier. A tree-organized neural network can be trained to determine path probabilities and issue driving pattern judgments, with the tree-organized neural network including a soft tree model and a hard decision leaf. One or more driving pattern judgments can be issued based on a probability of possible driving patterns derived from the modality-specific features.

An image classification apparatus includes an image segmentation module configured to segment a learning image into a plurality of segment images, a primary classification module configured to perform machine learning on a primary classifier using the plurality of segment images, and a secondary classification module configured to calculate a weight value combination for creating a secondary classification estimation value for the learning image from a plurality of primary classification estimation values generated by passing the plurality of segment images to the trained primary classifier, or a machine learning-based learning parameter.