Hong Kong Applied Science and Technology Research Institute Company, Limited

Zheng Long Li

Yi Ping Tse

Laifa Fang

Sung Ho Cheung

Chun Ming Chan

A visual content search and retrieval platform comprising an object detection model, a coarse-grained image classification model, and a selection of fine-grained image classification models. During inference, the object detection model locates and extracts a query object in a query image; the coarse-grained image classification model classifies the query object's meta-category; the fine-grained image classification model trained and optimized for the meta-category of the query object is selected; and the selected fine-grained image classification model searches and retrieves from an image database images of objects most similar to the query object. The fine-grained image classification model is implemented by a multi-focus fine-grained image search and retrieval apparatus comprising a class learning module learned to generate the query object's overall appearance query features; a local description module learned to generate the query object's local details query features; an outline description module learned to generate the query object's outline query features.

Fine-grained visual content search platform

The framework predicts water-quality data from observation data associated with raw features, e.g., rainfall. In the framework, an artificial neural network (ANN) computes the predicted water-quality data from feature data associated with impact features and derived from the observation data. The impact features are learnable, and are selected from impact-feature candidates comprising directional change- (DC-based) features each indicating occurrence of DC events in a corresponding raw feature. Including the DC-based features in the candidates enhances the ANN's ability of capturing significant change patterns of water quality due to extreme/unexpected events. The ANN architecture is also configurable according to model hyperparameters, which are learnable. The impact features and model hyperparameters are learnt by differential evolution for maximizing a prediction performance achieved by the ANN, thereby enabling the ANN architecture and impact features to be automatically optimized without requiring manual adjustment by domain experts in applying the ANN to different situations.

Growing community of inventors

Hong Kong, China

Zheng Long Li