Implicit Neural Representations (INRs) have revolutionized signal processing and computer vision by modeling signals as continuous, differentiable functions parameterized by neural networks. However, INRs are prone to the spectral bias problem, limiting their ability to retain high-frequency information, and often struggle with noise robustness. Motivated by recent trends in iterative refinement processes, we propose Iterative Implicit Neural Representations (I-INRs). This novel plug-and-play framework iteratively refines signal reconstructions to restore high-frequency details, improve noise robustness, and enhance generalization, ultimately delivering superior reconstruction quality. I-INRs integrate seamlessly into existing INR architectures with only a 0.5–2% increase in parameters. During reconstruction, the iterative refinement adds just 0.8–1.6% additional FLOPs over the baseline while delivering a substantial performance boost of up to +2.0 PSNR. Extensive experiments demonstrate that I-INRs consistently outperform WIRE, SIREN, and Gauss across various computer vision tasks, including image fitting, image denoising, and object occupancy prediction.

A significant volume of analog information, i.e., documents and images, have been digitized in the form of scanned copies for storing, sharing, and/or analyzing in the digital world. However, the quality of such contents is severely degraded by various distortions caused by printing, storing, and scanning processes in the physical world. Although restoring high-quality content from scanned copies has become an indispensable task for many products, it has not been systematically explored, and to the best of our knowledge, no public datasets are available. In this paper, we define this problem as Descanning and introduce a new high-quality and large-scale dataset named DESCAN-18K. It contains 18K pairs of original and scanned images collected in the wild containing multiple complex degradations. In order to eliminate such complex degradations, we propose a new image restoration model called DescanDiffusion consisting of a color encoder that corrects the global color degradation and a conditional denoising diffusion probabilistic model (DDPM) that removes local degradations. To further improve the generalization ability of DescanDiffusion, we also design a synthetic data generation scheme by reproducing prominent degradations in scanned images. We demonstrate that our DescanDiffusion outperforms other baselines including commercial restoration products, objectively and subjectively, via comprehensive experiments and analyses.

Food Computing is an emerging research field that leverages Natural Language Processing (NLP) techniques to extract valuable insights from textual data. A key task within NLP is Named Entity Recognition (NER), which involves identifying and categorizing words or phrases into predefined categories. Current, NER methods are limited in their capacity to recognize novel entity types, such as food names. Enhancing their capabilities to encompass new entities necessitates supervised training, that needs substantial labeled dataset. Labeling such datasets is time-intensive and challenging, particularly for novel entities like foods, that lack standardized definitions across various applications. Furthermore, existing state-of-the-art transformer-based techniques are not suitable for lightweight applications due to their large size and computational complexity. In this study, we present a neuro-heuristic based approach for food name recognition, specifically targeting food names or recipe names. To mitigate the need for extensive labeling, we adopt a template-based approach to prepare a dataset with labeled food entities. Our system achieves an impressive F1 accuracy of 0.97, on the dataset prepared by using multiple publicly available resources, including the Branded Food Dataset and NPR Dataset.

As an important active remote sensing tool in forest remote sensing, lidar is able to provide information on tree height, canopy structure, aboveground biomass, among other parameters. It has become desirable to be able to classify tree species using lidar data during recent years. Research has been performed using commercial non-polarimetric lidar in tree species classification, at either dominant species level or individual tree level. The objective of this research is to classify deciduous and coniferous trees using the newly developed polarimetric lidar system. Lidar data from five different tree species were collected in the field. These included ponderosa pine, Austrian pine, blue spruce, green ash and maple. Data were preprocessed and artificial neural network method was developed for classification. Data analysis demonstrated that the classification performance using polarimetric lidar data was far better than that using the non-polarimetric lidar data.