Relation-enhanced DETR for Component Detection in Graphic Design Reverse Engineering
Authors: Xixuan Hao, Danqing Huang, Jieru Lin, Chin-Yew Lin
IJCAI 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiment results on three public datasets show that our approach achieves better performance than several strong baselines. We conduct experiments on three publicly-available graphic design datasets, including RICO [Deka et al., 2017] (mobile UI), Crello [Yamaguchi, 2021] (posters) and Info PPT [Shi et al., 2022] (slides). |
| Researcher Affiliation | Collaboration | Xixuan Hao1 , Danqing Huang2 , Jieru Lin3 and Chin-Yew Lin2 1The University of Hong Kong 2Microsoft Research 3Harbin Institute of Technology hxxjxw@connect.hku.hk, {dahua, cyl}@microsoft.com, hitjierulin@gmail.com |
| Pseudocode | No | The paper describes the proposed method in text and with diagrams (Figure 4, Figure 5) but does not include explicit pseudocode or algorithm blocks. |
| Open Source Code | No | The paper states: 'We use open-source implementations of Faster-RCNN 2, Conditional DETR3 and DAB-DETR4.' followed by footnotes with URLs. However, this refers to third-party code used, not the authors' own code for their proposed method. |
| Open Datasets | Yes | We consider three publicly-available graphic design datasets: RICO, Crello and Info PPT1. (1) RICO [Deka et al., 2017] ... (2) Crello [Yamaguchi, 2021] ... (3) Info PPT [Shi et al., 2022] ... |
| Dataset Splits | No | The paper mentions the total number of images in the datasets but does not provide specific training, validation, or test split percentages or counts. It refers to 'standard metrics from COCO detection evaluation criteria' but not the data partitioning. |
| Hardware Specification | Yes | We run all the models on 8 Tesla V100 GPUs with batch size 8 for 40 epochs and AdamW [Loshchilov and Hutter, 2017] is used for training with weight decay 10−4. |
| Software Dependencies | No | The paper mentions using 'AdamW' and 'Cosine Annealing optimizer', and references 'open-source implementations of Faster-RCNN, Conditional DETR, and DAB-DETR', but does not provide specific version numbers for any software libraries, frameworks, or programming languages. |
| Experiment Setup | Yes | We run all the models on 8 Tesla V100 GPUs with batch size 8 for 40 epochs and AdamW [Loshchilov and Hutter, 2017] is used for training with weight decay 10−4. We set different learning rates for backbone and other modules to 10−5 and 10−4 respectively. Cosine Annealing optimizer is used with Tmax of 40 and decays it by a factor of 0.05 by the end of training. During training, images are resized such that the short side is at least 480 and at most 800 pixels and the long size is at most 1333 pixels. |