Upsampling

About: Upsampling is a research topic. Over the lifetime, 2426 publications have been published within this topic receiving 57613 citations.

Patterns from multiresolution 0-1 data

Abstract: Biological systems are complex systems and often the biological data is available in different resolutions. Computational algorithms are often designed to work with only specific resolution of data. Hence, upsampling or downsampling is necessary before the data can be fed to the algorithm. Moreover, high-resolution data incorporates significant amount of noise thus producing explosion of redundant patterns such as maximal frequent itemset, closed frequent itemset and non-derivable itemset in the data which can be solved by downsampling the data if the information loss is insignificant during sampling. Furthermore, comparing the results of an algorithm on data in different resolution can produce interesting results which aids in determining suitable resolution of data. In addition, experiments in different resolutions can be helpful in determining the appropriate resolution for computational methods. In this paper, three methods of downsampling are proposed, implemented and experiments are performed on different resolutions and the suitability of the proposed methods are validated and the results compared. Mixture models are trained on the data and the results are analyzed and it was seen that the proposed methods produce plausible results showing that the significant patterns in the data are retained in lower resolution. The proposed methods can be extensively used in integration of databases.

Encoding Contextual Information by Interlacing Transformer and Convolution for Remote Sensing Imagery Semantic Segmentation

Abstract: Contextual information plays a pivotal role in the semantic segmentation of remote sensing imagery (RSI) due to the imbalanced distributions and ubiquitous intra-class variants. The emergence of the transformer intrigues the revolution of vision tasks with its impressive scalability in establishing long-range dependencies. However, the local patterns, such as inherent structures and spatial details, are broken with the tokenization of the transformer. Therefore, the ICTNet is devised to confront the deficiencies mentioned above. Principally, ICTNet inherits the encoder–decoder architecture. First of all, Swin Transformer blocks (STBs) and convolution blocks (CBs) are deployed and interlaced, accompanied by encoded feature aggregation modules (EFAs) in the encoder stage. This design allows the network to learn the local patterns and distant dependencies and their interactions simultaneously. Moreover, multiple DUpsamplings (DUPs) followed by decoded feature aggregation modules (DFAs) form the decoder of ICTNet. Specifically, the transformation and upsampling loss are shrunken while recovering features. Together with the devised encoder and decoder, the well-rounded context is captured and contributes to the inference most. Extensive experiments are conducted on the ISPRS Vaihingen, Potsdam and DeepGlobe benchmarks. Quantitative and qualitative evaluations exhibit the competitive performance of ICTNet compared to mainstream and state-of-the-art methods. Additionally, the ablation study of DFA and DUP is implemented to validate the effects.

Reduce Information Loss in Transformers for Pluralistic Image Inpainting

Abstract: Transformers have achieved great success in pluralistic image inpainting recently. However, we find existing transformer based solutions regard each pixel as a token, thus suffer from information loss issue from two aspects: 1) They downsample the input image into much lower resolutions for efficiency consideration, incurring information loss and extra misalignment for the boundaries of masked regions. 2) They quantize 256 ³ RGB pixels to a small number (such as 512) of quantized pixels. The indices of quantized pixels are used as tokens for the inputs and prediction targets of transformer. Although an extra CNN network is used to upsample and refine the low-resolution results, it is difficult to retrieve the lost information back. To keep input information as much as possible, we propose a new transformer based framework “PUT”. Specifically, to avoid input downsampling while maintaining the computation efficiency, we design a patch-based auto-encoder P-VQVAE, where the encoder converts the masked image into non-overlapped patch tokens and the decoder recovers the masked regions from the inpainted tokens while keeping the unmasked regions unchanged. To eliminate the information loss caused by quantization, an Un-Quantized Transformer (UQ-Transformer) is applied, which directly takes the features from P-VQVAE encoder as input without quantization and regards the quantized tokens only as prediction targets. Extensive experiments show that PUT greatly outperforms state-of-the-art methods on image fidelity, especially for large masked regions and complex large-scale datasets.

Fast randomized multi-scale energy minimization for inferring depth from stereo image pairs

Abstract: An image processing module infers depth from a stereo image pair according to a multi-scale energy minimization process. A stereo image pair is progressively downsampled to generate a pyramid of downsampled image pairs of varying resolution. Starting with the coarsest downsampled image pair, a disparity map is generated that reflects displacement between corresponding pixels in the stereo image pair. The disparity map is then progressively upsampled. At each upsampling stage, the disparity labels are refined according to an energy function. The disparity labels provide depth information related to surfaces depicted in the stereo image pair.

BLAST-NET: Semantic Segmentation of Human Blastocyst Components via Cascaded Atrous Pyramid and Dense Progressive Upsampling

Abstract: Components of a human blastocyst (day-5 embryo) and their morphological attributes highly correlate with the embryo’s potentials for a viable pregnancy. Automatic semantic segmentation of human blastocyst components is a crucial step toward achieving objective quality assessment of such blastocyst. In this paper, a semantic segmentation system is proposed for human blastocyst components in microscopic images. The proposed Blast-Net features two novel components: a Cascaded Atrous Pyramid Pooling (CAPP) module to incorporate multi-scale global contextual priors, and a Dense Progressive Sub-pixel Upsampling (DPSU) module to recover the high-resolution prediction map. Experimental results confirm that the proposed method achieves the best-reported segmentation performance to date with a mean Jaccard Index of 82.85 % for microscopic images of the human blastocyst.