A small collected dataset for the workshop paper at ICCV-RLQ'2019.
https://drive.google.com/open?id=1WtO8yO0zsv8wu_m3CAEA_K0vn1lv9W7H
The folder includes all data of 5 categories BILL, CAPTION, (Visa) CARD, NAMECARD, and PRODUCT with their ground truths of text lines; and the detected results of 6 selected state-of-the-art methods: CRAFT, CTPN, EAST, FOTS, PixelLink, and PSENet.
The folder is organized as:
-hblab-textdetection --- images
|---------- groundtruth
|---------- exp-results ( all .txt and images)
|--------- CRAFT
|--------- CTPN
|--------- EAST
|--------- FOTS
|--------- PixelLink
|--------- PSENet
Please cite our work if you use our datasets or experimental results: Diep Thi Ngoc Nguyen, State-of-the-Art in Action: Unconstrained Text Detection, ICCV RLQ 2019