An Auto-Grading Oriented Approach for Off-Line Handwritten Organic Cyclic Compound Structure Formulas Recognition

Auto-grading, as an instruction tool, could reduce teachers’ workload, provide students with instant feedback and support highly personalized learning. Therefore, this topic attracts considerable attentions from researchers recently. To realize the automatic grading of handwritten chemistry assignments, problem chemical notations recognition should be solved first. The recent solutions belonging to end-to-end trainable category suffered lacking accurate alignment information between input output. They serve aim reading into electrical devices better prepare relevant e-documents instead auto-grading assignments. tackle limitation enable assignments at a fine-grained level. In work, we propose component-detection-based approach for recognizing off-line Organic Cyclic Compound Structure Formulas (OCCSFs). Specifically, define different components OCCSFs objects (including graphical text objects), adopt deep learning detector detect them. Then, regarding detected objects, introduce improved attention-based encoder-decoder model recognition. Finally, these detection results geometric relationships article designs holistic algorithm interpreting spatial structure OCCSFs. proposed method is evaluated on self-collected data set consisting 3000 samples achieves promising results.