黑色透明图 背景

“Lilou” Machine Vision Intelligence Innovation Platform

“Lilou” Machine Vision Intelligence Innovation Platform is a platform for research and development, training, releasing and deployment of intelligent models and algorithms for the Industrial Internet.

1 /8
Introduction to Products

“Lilou” Machine Vision Intelligence Innovation Platform is a platform for research and development, training, releasing and deployment of intelligent models and algorithms for the Industrial Internet. It is tailored for Industrial Internet, aiming to solve pain points such as batch quality problems in large-scale production, low automation levels in traditional factories, continuously rising labor costs, and low efficiency in production processes and ex-factory manual inspections.
It has technological advantages such as small sample capability, knowledge distillation capability, lightweight model capability, and natural adversarial training capability, while also integrating research, training, release, and deployment procedures. Therefore, users can easily create and train their own intelligent models through this platform. It also provides a rich library of models and algorithms, supporting training and debugging of mainstream algorithm models such as CNN, RNN, LSTM, ATTENTION, etc, and supporting a variety of application scenarios, such as natural language processing, image processing, and language recognition. Besides, it supports functions such as autonomous hyperparameter search, data augmentation, and model optimization, helping users quickly build efficient intelligent models. With a variety of APIs and SDKs, it also helps customers easily integrate intelligent technology into their own applications, providing them with end-to-end overall solutions.

Product Features
Small Sample Capability Small Sample Capability Lightweight Capability
A non-negative implicit semantic analysis model based on instance frequency weighted regularization strategy is set up to effectively improve the generalization ability of the model, and significantly reduce the training sample size. A non-negative implicit semantic analysis model based on instance frequency weighted regularization strategy is set up to effectively improve the generalization ability of the model, and significantly reduce the training sample size. The construction of technical features is based on data density-oriented modeling theory and embedded lightweight technology, which maintains the robustness and representation accuracy of the model with 80% reduction in training time.

×
Contact us
Official Customer Service: +86-531-81604000-9
E-mail : servicecenter@inspur.com