Monday, November 19, 2018
The AI models can be used for a variety of image recognition tasks, such as moderating user content (e.g., detecting nudity or guns), tagging images with visible ads, detecting damaged products, classifying x-ray photos and much more. SentiSight.ai has applications in e-commerce, industry, agriculture, medicine and beyond.
"With SentiSight.ai, you can quickly and easily implement modern deep learning techniques to build an AI model and immediately start identifying content in images," said Dr. Karolis Uziela, SentiSight.ai team lead from Neurotechnology. "It enables users to manage a tremendous number of images in an easy and efficient way."
SentiSight.ai provides the tools users need to complete the full process of creating a model, including:
Image labeling toolkit. The process of data labeling is very important, but it can also be time-consuming. SentiSight.ai tools make this work easier and a lot more efficient. Image labeling is done using an intuitive interface and the output labels are automatically saved in a correct format that is suitable for deep learning algorithms.
Interactive model training environment. Interactive model training provides the ability to track training processes and explore the results. SentiSight.ai produces statistics—prediction accuracy, precision, recall, F1 score and many others—that allow users to measure their models' performance. Users can immediately view, filter and download the predictions. Trained models can either be used online inside the SentiSight.ai platform or via REST API.
Custom projects. For those who have a large and complicated project or just need a little more help, Neurotechnology experts can assist with the workflow.
SentiSight.ai replaces SentiSight SDK in Neurotechnology's product line. The new version employs cutting-edge, deep learning algorithms that update the previous version's more traditional approach to image recognition. SentiSight SDK could only recognize particular learned objects and not object classes. Next-generation SentiSight.ai is able to identify classes of objects that are more abstract and can differ quite a lot from each other.
For more information about SentiSight.ai, please visit: www.sentisight.ai.