Knowledge_extraction_to_lean
Nightime image comparison
150_evframe,164_evframe,567_spalling : good crack , in image invisible to detect.
504 frame,604 evframe - too much light crack invisible in evframe, but visible in evframe. 490 frame,608 evframe - fast movement in low light.
** In our yolov6 training we use achor-free concept.Also we keep the Exponential moving average (EMA) of our model weights (self.ema = ModelEMA(model) if self.main_process else None) to improve the model stability and the generalizability.
Papers
- Direct training high-performance spiking neural networks for object recognition and detection** https://www.frontiersin.org/journals/neuroscience/articles/10.3389/fnins.2023.1229951/full