Release the edge AI image processing model, which can be converted into low noise image in real time.

Output Image based on LeapMind minimal quantization Model

　On January 31, 2022, LeapMind released the AI (artificial Intelligence) Image processing Model, which can be run on edge devices. So far, this has been a problem for edge devices to allow real-time operation and high-quality image quality processing, and noisy images can be improved to the same image quality as smartphones. The evaluation version has been available since February of the same year. The model uses LeapMind's ultra-low power AI inference accelerator IP "Efficiera (Efficiera)" to provide Raw-to-Raw noise reduction model, noise reduction processing based on deep learning, and relearnable learned model. In addition, in addition to the performance scalability of Efficiera, which can shoot animation in real time, the company's "Pixel embedding technology" and "minimization technology" make it possible to achieve high performance, high picture quality and lightweight. With these features, AI can improve image quality even without high-cost high-sensitivity sensors and large lenses, providing image quality no less than traditional 32-bit floating-point models. In addition, even the moving image cameras used in low illumination and low exposure conditions, such as surveillance cameras and inspection cameras, can be expected to improve the accuracy of object recognition and inspection.

MONOist (male)