Deep-learning-based nanowire detection in AFM images for automated nanomanipulation

Atomic force microscope (AFM)-based nanomanipulation has been proved to be a possible method for assembling various nanoparticles into complex patterns and devices. To achieve efficient fully automated nanomanipulation, on the substrate must identified precisely automatically. This work focuses an autodetection flexible nanowires using deep learning technique. An instance segmentation network based You Only Look Once version 3 (YOLOv3) convolutional (FCN) is applied segment all movable in AFM images. Combined with follow-up image morphology fitting algorithms, this enables detection of postures positions at high abstraction level. Benefitting from these our program able automatically detect different morphologies nanometer resolution over 90% reliability testing dataset. The results are less affected by complexity than existing methods demonstrate good robustness algorithm.