Quantification of data including micrographs, which was difficult until now, is possible.
In addition to subjective judgments by human beings on research results, credibility can be improved by adding objective judgments using AI.
In addition, being able to make objective judgments can prevent wrong judgments due to cognitive bias. This also saves research costs.
How did AI learn and how did it come up with answers? Visualization is a technology that makes this easy for humans to understand. It is possible to visualize how the AI is identifying the object, and by visualizing the output of each layer, you can see how it reacts to the input as an image.
At deeper levels, more complex and global features are learned.
The figure is a visualization of the fifth layer of the CNN pooling layer.
The dark part is reacting strongly. AI judges this image using the darker part as the criterion.
Input is an image of a person. From this visualized figure, you can see that AI pays attention to the silhouette of the head and neck and the face. (Red frame)
In our system, AI can be installed on the cloud, so it can flexibly support various applications.
Even when specially designed AI is used, it can be made compact without sacrificing accuracy by bit reduction technology. As a result, the size of AI and the time required for computation can be reduced to a fraction of the cost, greatly reducing costs.
In addition to producing products for companies, custom AI for specific research can be developed and operated at a low price. Please consider our proposal.