Multi-modal In-Context Learning Makes an Ego-evolving Scene Text Recognizer | IEEE Conference Publication | IEEE Xplore