Arbitrary-Oriented Scene Text Detection via Rotation Proposals

By Jianqi Ma, Weiyuan Shao, Hao Ye, Li Wang, Hong Wang, Yingbin Zheng, Xiangyang Xue. 2017

1,317 citations recorded by the source corpus.

This paper introduces a novel rotation-based framework for arbitrary-oriented text detection in natural scene images. We present the Rotation Region Proposal Networks, which are designed to generate inclined proposals with text orientation angle information. The angle information is then adapted for bounding box regression to make the proposals more accurately fit into the text region in terms of the orientation. The Rotation Region-of-Interest pooling layer is proposed to project arbitrary-oriented proposals to a feature map for a text region classifier. The whole framework is built upon a region-proposal-based architecture, which ensures the computational efficiency of the arbitrary-oriented text detection compared with previous text detection systems. We conduct experiments using the rotation-based framework on three real-world scene text detection datasets and demonstrate its superiority in terms of effectiveness and efficiency over previous approaches.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

1. Dacheng Tao
2. Luc Van Gool
3. Chunhua Shen
4. Xiang Bai
5. Yu Qiao
6. Ming-Hsuan Yang
7. Xiaogang Wang
8. Yi Yang

Citations in the corpus

• All You Need Is Boundary: Toward Arbitrary-Shaped Text Spotting (2019, methodology): “L or is the loss that regression from axis-aligned rectangular proposal to oriented rectangular box, which is similar as (Ma et al. 2018).”
• Instance Segmentation for Autonomous Log Grasping in Forestry Operations (2022, methodology): “The only changes applied to the rotated version of Mask RCNN are the use of RRPN and RRoIAlign [10] for oriented region proposals.”
• Improving Rotated Text Detection with Rotation Region Proposal Networks (2018, methodology): “In the RRPN paper [2], the authors extended RoI Pooling to RRoI Pooling by applying rotation transformation.”
• PIoU Loss: Towards Accurate Oriented Object Detection in Complex Environments (2020, neutral): “7 <1fps RRPN [28] ResNet101 800 × 800 × 79.”
• MPDIoU: A Loss for Efficient and Accurate Bounding Box Regression (2023, methodology): “There are other detectors, such as RRPN [24] and R2CNN [25] adding rotation angle regression to detect arbitrary-orientated objects for remote sensing detection and scene text detection.”
• Shape Robust Text Detection With Progressive Scale Expansion Network (2018, methodology): “2 RRPN [28] - 82.”
• ICDAR2017 Robust Reading Challenge on Multi-Lingual Scene Text Detection and Script Identification - RRC-MLT (2017, background): “2) CAS & Fudan University: Authors: Jianqi Ma, Weiyuan Shao, Yingbin Zheng, Hong Wang, Li Wang, Hao Ye, Xiangyang Xue Affiliation: Shanghai Advanced Research Institute, CAS & Fudan University Method name: SARI FDU RRPN v1 Method description: A Rotation Region Proposal Network (RRPN) [8] is designed ”
• Scene Text Detection and Recognition: The Deep Learning Era (2018, methodology): “Similarly, rotating region proposals [90] are generated with 6 different orientations.”
• Real-time Scene Text Detection with Differentiable Binarization (2019, neutral): “1 RRPN (Ma et al. 2018) 82 68 74 RRD (Liao et al.”
• Dense Label Encoding for Boundary Discontinuity Free Rotation Detection (2020, methodology): “RRPN [26], TextBoxes++ [18] and RRD [19] and FOTS [23] are some advanced methods for scene text detection.”
• Efficient and Accurate Arbitrary-Shaped Text Detection With Pixel Aggregation Network (2019, methodology): “Based on Faster R-CNN, RRPN [38] develops rotated region proposals to detect titled text.”
• Object Detection in Aerial Images: A Large-Scale Benchmark and Challenges (2021, neutral): “We also implemented modules such as rotated RoI Align and rotated position-sensitive RoI Align for rotated region feature extraction, which are crucial components in algorithms such as rotated region proposal network (RRPN) [89] and RoI Transformer [18].”
• FOTS: Fast Oriented Text Spotting with a Unified Network (2018, methodology): “We also compare to RRoI pooling proposed in RRPN [39].”
• Real-Time Scene Text Detection With Differentiable Binarization and Adaptive Scale Fusion (2022, neutral): “1 RRPN [40] 82 68 74 RRD [27] 87 73 79 10 MCN [33] 88 79 83 PixelLink [10] 83 73.”
• An Empirical Study of Remote Sensing Pretraining (2022, methodology): “Network ViTAE-S [14] ViTAEv2-S [28] Stage 3 4 Downsampling Ratio [4, 2, 2] [4, 2, 2, 2] Embedding Dim [64, 64, 192] [64, 64, 128, 256] Stage Dim [96, 192, 384] [64, 128, 256, 512]”
• Sliding Line Point Regression for Shape Robust Scene Text Detection (2018, background): “Considering the particularity of text line, [26] appended different angle anchors which are suitable for arbitrary-oriented text line.”
• RBox-CNN: rotated bounding box based CNN for ship detection in remote sensing image (2018, methodology): “[8] proposed Rotation Region Proposal Networks (RRPN) based on Faster R-CNN.”