Description: After viewing training data, we observe that the text inside is small and maybe rectangle box is enough for it. So we use a yolox detector based tracker to capture text.