Patchy Image Structure Classification Using Multi-Orientation Region Transform
Authors: Xiaohan Yu, Yang Zhao, Yongsheng Gao, Shengwu Xiong, Xiaohui Yuan12741-12748
AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Very encouraging experimental results on the challenging ultra-fine-grained cultivar recognition task, insect wing recognition task, and large variation butterfly recognition task are obtained, which demonstrate the effectiveness and superiority of the proposed MORT over the state-of-the-art methods in classifying patchy image structures. |
| Researcher Affiliation | Academia | Xiaohan Yu,1,2 Yang Zhao,1,2 Yongsheng Gao,1, Shengwu Xiong,2,* Xiaohui Yuan2 1School of Engineering and Built Environment, Griffith University, Australia 2School of Computer Science and Technology, Wuhan University of Technology, China |
| Pseudocode | Yes | Algorithm 1 Calculating discrete Multi-Orientation Region Transform |
| Open Source Code | Yes | Our code and three patchy image structure datasets are available at: https://github.com/XiaohanYu-GU/MReT2019. |
| Open Datasets | Yes | Our code and three patchy image structure datasets are available at: https://github.com/XiaohanYu-GU/MReT2019. The Soy Cultivar Vein dataset is a publicly available dataset, which comprises 100 categories (cultivars) with 6 samples (leaf images) in each cultivar and thus has a total number of 100 6 = 600 images (Yu et al. 2019). The butterfly patchy image structure dataset (Btf PIS), is constructed by applying the canny edge detection (Canny 1986) to the binarized images of the first 50 images in each class from the public available Leeds butterfly dataset (Wang, Markert, and Everingham 2009). The insect wing patchy image structure dataset (Iw PIS) (Tofilski 2004), is adopted for evaluation, which comprises 25 classes of insect wings with 2 samples in each class. |
| Dataset Splits | No | For Soy Cultivar Vein, Btf PIS and Iw PIS datasets, we select the first half images from each category as the training set, and the remaining images as the testing set. The paper describes a training and testing split, but does not explicitly mention a separate validation split for hyperparameter tuning or early stopping. |
| Hardware Specification | No | The paper mentions that models are implemented in PyTorch and fine-tuned, but it does not provide any specific hardware details such as GPU models, CPU specifications, or memory. |
| Software Dependencies | Yes | All the models are implemented in Pytorch 1.0.0 |
| Experiment Setup | Yes | In all experiments, the input images are resized to 440 440, and cropped to 384 384 randomly for training. Standard data augmentations are applied including random rotation within 15 degree and horizontal flip with 0.5 probability. For fast MPN-COV and improved B-CNN, the models are trained with the best setting in the implementation of fast-MPNCOV (Li et al. 2018). For the remaining methods, the models are trained with the default settings in the implementation of DCL (Chen et al. 2019). |