These papers(about computer vision) are discussed in our research group in 2019 Fall. Our main research is mathematical expressions recognition and insect images recognition.
Track, Attend and Parse (TAP): An End-to-end Framework for Online Handwritten Mathematical Expression Recognition
In this paper, the seq2seq method using guided hybrid attention(GHA) is used to recognize mathematical expressions. They have achieved the state-of-the-art accuracy of more than 60%, which can not be defeated by the traditional methods like sequential methods.
In this paper, the seq2seq domain adaptation network for robust text image recognition is built, towards real world applications in various recognition scenarios, including the natural scene text, handwritten text and even mathematical expression using Gated Attention Similarity(GAS) unit.
In this paper, Region Decomposition and Assembly Detector Network Structure is built for object detection. This network is built upon the faster R-CNN without using feature pyramid, and can achieve a high accuracy even there is object occlusion. We have discussed this paper because object detection can help us to recognize mathematical expressions and insect images better.