Overview

What is EPicker

EPicker is a standalone software for particle picking in cryo-EM. It is implemented based on CenterNet, an open-source deep learning detection framework. Trained and well tested on datasets containing heterogenous particles, EPicker is able to pick particles, vesicles and fibers effectively.

Different from other particle picking tools, EPicker supports a continual learning mechanism. With a trained model and corresponding exemplar dataset generated from the old datasets, EPicker builds up new knowledge on the old model through a continual training process. The new model yielding from the continual learning is able to recognize both new particles and old ones. The continual learning avoids catastrophic forgetting known in fine-tuning method, and avoids joint training on multiple datasets by dividing a single training to multi-time training. With these advantages, EPikcer can continually upgrade the model to adapt more new features during the routine applicaitons with minor training time.

About this guide

This user guide includes all instructions you may need for downloading, installation, picking and training. A tutorial is also provided at the end of this guide to help you go through the complete workflow. If any questions or bugs, please contact the team through email: thuem2018@126.com