Distortion-aware CNNs for Spherical Images

Authors: Qiang Zhao, Chen Zhu, Feng Dai, Yike Ma, Guoqing Jin, Yongdong Zhang

IJCAI 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental To evaluate our method, we apply our network in spherical image classification problems based on transformed MNIST and CIFAR-10 datasets. Compared with the baseline method, our method can get much better performance. We also analyze the variants of our network.
Researcher Affiliation Academia 1 Key Laboratory of Intelligent Information Processing of Chinese Academy of Sciences, Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China 2 University of Chinese Academy of Sciences, Beijing, China 3 University of Science and Technology of China, Hefei, China
Pseudocode No The paper does not include any sections or figures explicitly labeled as 'Pseudocode' or 'Algorithm'.
Open Source Code No The paper does not provide any statements about releasing code, nor does it include links to a code repository.
Open Datasets Yes To evaluate our method, we transform the well known MNIST and CIFAR-10 dataset to spherical ones and compare our method with baseline method... The CIFAR-10 dataset consists of 50,000 training images and 10,000 testing images in 10 classes... MNIST is a dataset of handwritten digits, which has a training set of 60,000 images and a test set of 10,000 images.
Dataset Splits No The paper mentions training and test sets but does not explicitly specify a separate validation dataset split or how validation was performed beyond general training parameters.
Hardware Specification No The paper mentions that Caffe is 'efficiently computed on GPU devices' and that the GPU version uses 'CUDA kernels' but does not specify any particular GPU model (e.g., NVIDIA V100, RTX 3090) or other hardware components (CPU, RAM).
Software Dependencies No We implement our distortion-aware CNNs based on Caffe framework [Jia et al., 2014]. When training the networks, we use ADAM algorithm [Kingma and Ba, 2014] with minibatch size of 128. While Caffe and ADAM are mentioned, specific version numbers for Caffe or any other libraries are not provided.
Experiment Setup Yes When training the networks, we use ADAM algorithm [Kingma and Ba, 2014] with minibatch size of 128. The learning rate is set as 10e-3. For networks containing 5 modules, the size of feature map in each module is 32, 32, 64, 64 and 64 respectively. The network is ended with a 10-way fullyconnected layer and softmax.