RRL: Regional Rotate Layer in Convolutional Neural Networks

Authors: Zongbo Hao, Tao Zhang, Mingwang Chen, Zou Kaixu826-833

AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Evaluate our module with Le Net-5, Res Net-18 and tiny-yolov3, we get impressive results.
Researcher Affiliation Academia Zongbo Hao, Tao Zhang, Mingwang Chen, Zou Kaixu University of Electronic Science and Technology of China No.4, Section 2, North Jianshe Road Chengdu, China, 610054 zbhao@uestc.edu.cn, zhangtao897476472@126.com, chenmingwang815@163.com, 1241510971@qq.com
Pseudocode Yes Algorithm 1: RRL in local windows Input: RGB image sample batch {I1, I2, , It } Output: Rotate the feature maps to the same state
Open Source Code No The paper does not provide any links to open-source code for the described methodology.
Open Datasets Yes CIFAR-10 is used in our experiment. The dataset was proposed by krizhevsky in 2009. It contains 60000 32 32 colour images, belonging to 10 categories. There are 50000 images in training set (5000 in each category) and 10000 images in test set (1000 in each category). The plankton dataset(Cowen et al. 2015) consists of 30,336 gray images of different sizes, which are unevenly divided into 121 categories. Pascal VOC dataset contains 20 categories. The dataset has been widely used in object detection, semantic segmentation and classification tasks, and as a common test benchmark. VOC 2007 and VOC 2012 are used in this experiment.
Dataset Splits No The paper specifies training and testing splits for datasets like CIFAR-10 and Plankton, but does not explicitly mention or detail a separate validation set split for hyperparameter tuning or model selection.
Hardware Specification No The paper does not provide any specific details about the hardware used for running the experiments (e.g., GPU models, CPU types, memory).
Software Dependencies No The paper does not specify any software dependencies with version numbers (e.g., programming languages, libraries, frameworks).
Experiment Setup No The paper mentions '100,000 epochs of training' for the Plankton dataset, but it lacks specific details on other key hyperparameters (e.g., learning rate, batch size, optimizer) or system-level training configurations for any of the models (LeNet-5, ResNet-18, tiny-yolov3).