Multi-scale Resolution Research Articles

Self-Supervised Object Distance Estimation Using a Monocular Camera.

Distance estimation using a monocular camera is one of the most classic tasks for computer vision. Current monocular distance estimating methods need a lot of data collection or they produce imprecise results. In this paper, we propose a network for both object detection and distance estimation. A network-based on ShuffleNet and YOLO is used to detect an object, and a self-supervised learning network is used to estimate distance. We calibrated the camera, and the calibrated parameters were integrated into the overall network. We also analyzed the parameter variation of the camera pose. Further, a multi-scale resolution is applied to improve estimation accuracy by enriching the expression ability of depth information. We validated the results of object detection and distance estimation on the KITTI dataset and demonstrated that our approach is efficient and accurate. Finally, we construct a dataset and conduct similar experiments to verify the generality of the network in other scenarios. The results show that our proposed methods outperform alternative approaches on object-specific distance estimation.

Cryo-ET of Toxoplasma parasites gives subnanometer insight into tubulin-based structures

Tubulin is a conserved protein that polymerizes into different forms of filamentous structures in Toxoplasma gondii, an obligate intracellular parasite in the phylum Apicomplexa. Two key tubulin-containing cytoskeletal components are subpellicular microtubules (SPMTs) and conoid fibrils (CFs). The SPMTs help maintain shape and gliding motility, while the CFs are implicated in invasion. Here, we use cryogenic electron tomography to determine the molecular structures of the SPMTs and CFs in vitrified intact and detergent-extracted parasites. Subvolume densities from detergent-extracted parasites yielded averaged density maps at subnanometer resolutions, and these were related back to their architecture in situ. An intralumenal spiral lines the interior of the 13-protofilament SPMTs, revealing a preferred orientation of these microtubules relative to the parasite's long axis. Each CF is composed of nine tubulin protofilaments that display a comma-shaped cross-section, plus additional associated components. Conoid protrusion, a crucial step in invasion, is associated with an altered pitch of each CF. The use of basic building blocks of protofilaments and different accessory proteins in one organism illustrates the versatility of tubulin to form two distinct types of assemblies, SPMTs and CFs.

Investigating Sea Surface Temperature and Coral Bleaching in the Coastal Area of Khanh Hoa Province

Khanh Hoa Coastal area is considered the area with the most diverse coral reefs in the west of the East Sea. With the trend of increasing global temperature, the coral reefs here are affected, including the phenomenon of bleaching. This paper uses a Multi-scale Ultra-high Resolution Sea Surface Temperature (MUR SST) data source to evaluate the possible relationship between sea surface temperature and coral bleaching in the period 2010-2019 in Khanh Hoa coastal area is based on two parameters: Hot Spot (HS) and Degree Heating Months (DHM). Research results show that in the past 10 years, corals in Khanh Hoa coastal area may experience heat stress in 6 years, including 2010, 2013 and the years from 2016 to 2019. The phenomenon of heat stress starts to occur in May of each year, and the level of heat stress in 2010 was stronger than in other years. Within 6 years, there have been heat stress, coral bleaching events due to temperature only occurred in 2010; while in 2013, 2016, 2017, 2018 and 2019, corals suffered from heat stress in the watch level, meaning an increase in temperature is not yet capable of causing coral bleaching. The cause of coral bleaching in 2010 was the combined effect of the increase in sea surface temperature and the suppression of upwelling during the southwest monsoon.

U_EFF_NET: High-Precision Segmentation of Offshore Farms From High-Resolution SAR Remote Sensing Images

Offshore aquaculture promotes the development of aquaculture industry and brings huge economic benefits to fishermen, while seriously affecting the near-coast environment. Accurate access to the range of offshore farms at home and abroad is of great significance for marine disaster warning and coastal management. Remote sensing is a very effective means of observing offshore farms. Offshore farms segmentation technology is more mature in high-resolution optical images. SAR images have the advantage of being available all day and all night. Using SAR images to extract offshore farms has become a recent research hotspot. Gaofen-3 and HISEA-1 satellites have high-resolution marine observation capabilities. Combined with their super observation capabilities, this paper proposes a technical solution to extract offshore farms using Gaofen-3 and HISEA-1 imags. Firstly, the combine-crop component module is proposed to create a new SAR images dataset based the original dataset. Next, common data augmentation methods such as flip, rotate, and copy-paste are used, the online multi-scale resolution image expansion is proposed to enrich the dataset and enlarge the data volume. Finally, the U_EFF_NET model is proposed. The model uses encoder-decoder structure and the light-weight feature extraction network EfficientNet-b0 as backbone. The decoder embeds the attention mechanism. The optimization strategy uses joint loss function to achieve multi-task learning, and the OHEM strategy is added to the joint loss function. The method proposed in this paper has high accuracy and high practicality as the frequency-weight intersection ratio of the extracted results on the test set is 98.12%. And the inference time is at the forefront.

Ocean Front Detection with Glider and Satellite-Derived SST Data in the Southern California Current System

This study proposes a method to detect ocean fronts from in situ temperature and density glider measurements. This method is applied to data collected along the CalCOFI Line 90, south of the California Current System (CCS), over the 2006–2013 period. It is based on image-processing techniques commonly applied to sea surface temperature (SST) satellite data. Front detection results using glider data are consistent with those obtained in other studies carried out in the CCS. SST images of the Multi-scale Ultra-high Resolution (MUR) dataset were also used to compare the probability of occurrence or front frequency (FF) obtained with the two datasets. Glider and MUR temperatures are highly correlated. Along Line 90, frontal frequency exhibited the same maxima near the transition zone (~130 km offshore) as derived from MUR and glider datasets. However, marked differences were found in the bimonthly FF probability with high (low) front frequency in spring-summer for glider (MUR) data. Methodological differences explaining these contrasting results are investigated. Thermohaline-compensated fronts are more abundant towards the oceanic zone, although most fronts are detected using both temperature and density criteria, indicating a significant contribution of temperature to density in this region.

Correlative Image-Based Release Prediction and 3D Microstructure Characterization for a Long Acting Parenteral Implant.

Imaging-based characterization of polymeric drug-eluting implants can be challenging due to the microstructural complexity and scale of dispersed drug domains and polymer matrix. The typical evaluation via real-time (and accelerated in vitro experiments not only can be very labor intensive since implants are designed to last for 3months or longer, but also fails to elucidate the impact of the internal microstructure on the implant release rate. A novel characterization technique, combining multi-scale high resolution three-dimensional imaging, was developed for a mechanistic understanding of the impact of formulation and manufacturing process on the implant microstructure. Artificial intelligence-based image segmentation and imaging analytics convert "visualized" structural properties into numerical models, which can be used to calculate key parameters governing drug transport in the polymer matrix, such as effective permeability. Simulations of drug transport in structures constructed on the basis of image analytics can be used to predict the release rates for the drug-eluting implant without running lengthy experiments. Multi-scale imaging approach and image-based characterization generate a large amount of quantitative structural information that are difficult to obtain experimentally. The direct-imaging based analytics and simulation is a powerful tool and has potential to advance fundamental understanding of drug release mechanism and the development of robust drug-eluting implants.

An Effective Multi-Task Two-Stage Network with the Cross-Scale Training Strategy for Multi-Scale Image Super Resolution

Convolutional neural networks and the per-pixel loss function have shown their potential to be the best combination for super-resolving severely degraded images. However, there are still challenges, such as the massive number of parameters requiring prohibitive memory and vast computing and storage resources as well as time-consuming training and testing. What is more, the per-pixel loss measured by L2 and the Peak Signal-to-Noise Ratio do not correlate well with human perception of image quality, since L2 simply does not capture the intricate characteristics of human visual systems. To address these issues, we propose an effective two-stage hourglass network with multi-task co-optimization, which enables the entire network to focus on training and testing time and inherent image patterns such as local luminance, contrast, structure and data distribution. Moreover, to avoid overwhelming memory overheads, our model is capable of performing real-time single image multi-scale super-resolution, so it is memory-friendly, meaning that memory space is utilized efficiently. In addition, in order to best use the underlying structure and perception of image quality and the intermediate estimates during the inference process, we introduce a cross-scale training strategy with 2×, 3× and 4× image super-resolution. This effective multi-task two-stage network with the cross-scale strategy for multi-scale image super-resolution is named EMTCM. Quantitative and qualitative experiment results show that the proposed EMTCM network outperforms state-of-the-art methods in recovering high-quality images.

Macaque neuron instance segmentation only with point annotations based on multiscale fully convolutional regression neural network

In the field of biomedicine, instance segmentation / individualization is important in analyzing the number, the morphology and the distribution of neurons for the whole slide images. Traditionally, biologists apply the stereology technique to manually count the number of neurons in the regions of interest and estimate the number in anatomical regions or the entire brain. This is very tedious and time-consuming. In this paper, we propose a multiscale fully convolutional regression neural network combined with a competitive region growing technique to individualize size-varying and touching neurons in the major anatomical regions of the macaque brain. Given that neuron instance or contour annotations are infeasible to obtain in certain regions, such as the dentate gyrus where thousands of touching neurons are present, we ask an expert to perform point annotations in the center location of neurons (noted as neuron centroids) for training. Thanks to the multiscale resolution achieved by parallel multiple receptive fields and different network depths, our proposed network succeeds in detecting the centroids of size-varying and touching neurons. Competitive region growing is then applied on these centroids to achieve neuron instance segmentation. Experiments on the macaque brain data suggest that our proposed method outperform the state-of-the-art methods in terms of neuron instance segmentation performance. To our knowledge, this is the first deep learning research work to individualize size-varying and touching neurons only using point annotations in major anatomical regions of the macaque brain.

The 3M Concept: Biomedical Translational Imaging from Molecules to Mouse to Man

Abstract Imaging keeps pervading biomedical sciences from the nanoscale to the bedside. Connecting the hierarchical levels of biomedicine with relevant imaging approaches, however, remains a challenge. Here we present a concept, called “3M”, which can deliver a question, formulated at the bedside, across the wide-ranging hierarchical organization of the living organism, from the molecular level, through the small-animal scale, to whole-body human functional imaging. We present an example of nanoparticle development pipeline extending from atomic force microscopy to pre-clinical whole body imaging methods to highlight the essential features of the 3M concept, which integrates multi-scale resolution and quantification into a single logical process. Using the nanoscale to human clinical whole body approach, we present the successful development, characterisation and application of Prussian Blue nanoparticles for a variety of imaging modalities, extending it to isotope payload quantification and shape-biodistribution relationships. The translation of an idea from the bedside to the molecular level and back requires a set of novel combinatorial imaging methodologies interconnected into a logical pipeline. The proposed integrative molecules-to-mouse-to-man (3M) approach offers a promising, clinically oriented toolkit that lends the prospect of obtaining an ever-increasing amount of correlated information from as small a voxel of the human body as possible.

Financion: A quantum approach to financial market modelling

In this paper we will show that, in the financial market, the concept of chaos is removed and replaced with that of complexity. Indeed, it has always shown the characteristics of a dynamical system with high complexity. In particular, we present a new theory to describe the financial phenomenologies: a theory that is able to combine the different founding aspects of the financial market. This study provides a useful answer to the fundamental aspects of the market: its multiscale or multi resolution properties; its quantum-relativistic properties; the usefulness of describing, at a fundamental level, the interactions between bullish o + and bearish o – financial operators; the usefulness of declining financial interactions between operators in terms of scattering, annihilation or creation of elementary liquidity constituents; the introduction of a specific field, named financial field, and its gauge boson, the so-called “financion” as a mediator of financial interactions; the possibility of describing the financial market as a dynamical system, capable of absorbing financions and that is quanta of interaction liquidity. Thanks to this new paradigm, we have been able to describe and define the market scale length, i.e. revenue, price, risk (λ, ρ, σ). Furthermore, we have analyzed Kahneman and Tversky’s Prospect Theory to describe the decision-making process under risk conditions, not considering the price as a function of time, p = p(t), but by carrying out the analysis of price dynamics as trajectories in the transformed energy-entropy space ES of the financial market , where there are more stable and reliable indications than common indicators.

Technical Note: Compact thermoacoustic imaging system based on a low-cost and miniaturized microwave generator for invivo biomedical imaging.

Most of existing thermoacoustic imaging (TAI) studies generally utilized a linear modulator-based high peak power magnetron generator (MG) for efficient TA signal excitation. However, a linear modulator-based MG is bulky and expensive. Here we present a low-cost and compact thermoacoustic imaging (TAI) system based on a miniaturized MG. The MG is based on solid-state modulator and operates at 3.05±0.025GHz, with a peak power of up to 60kW and adjustable pulse duration from 70 to 600ns. The dimensions and weight of this MG are 350×210×70mm3 (Width×Length×Height) and 7.5kg, respectively. The spatial resolution of the miniaturized MG-based TAI system is determined to be from 0.3 to 1.4mm using controlled phantom experiments. The system is further evaluated using invivo experiments where the finger joints and vasculature in the forearm and opisthenar of human participants are successfully imaged. This study suggests that the miniaturized MG based TAI systems can be used for invivo joint and vascular imaging with multiscale resolutions (0.3-1.4mm).

Cluster Identification Using Modularity Optimization to Uncover Chemical Heterogeneity in Complex Solutions.

Structural heterogeneity is commonly manifested in solutions and liquids that feature competition of different interparticle forces. Identifying and characterizing heterogeneity across different length scales requires multimodal experimental measurement and/or the application of new techniques for the interrogation of atomistic simulation data. Within the latter, the parsing of networks of interparticle interactions (chemical networks) has been demonstrated to be a valuable tool for identifying subensembles of chemical environments. However, chemical networks can adopt a wide variety of topologies that challenge generalizable methods for identifying heterogeneous behavior, and few network analysis algorithms have been proposed for multiscale resolution. In this study, we apply a method of partitioning using the graph theoretic concept of clusters and communities. Using a modularity optimization algorithm, the cluster partition creates subgraphs based on their relative internal and external connectivities. The methodology is tested on two soft matter systems that have significantly different network topologies so as to probe its ability to identify multiple scale features and its generalizability. A binary Lennard-Jones fluid is first examined, where one component causes subgraphs that have high internal network connectivity yet are still connected to the rest of the interparticle network of interactions. The impact of connectivity and edge weighting on the cluster partition is investigated. In the second system, hierarchically organized molecular structures comprised of hydrogen bonded water molecules are identified at a liquid/liquid interface. These structures have a much more sparse network with significantly varied internal connectivity that is a challenge to differentiate from the background hydrogen bonding network of water molecules at the instantaneous interface. The organized macrostructures are effectively isolated from the background network using the cluster partition, and a time-dependent implementation allows us to reveal their reactivity. These studies indicate that cluster partitioning based upon intermolecular network connectivity patterns is broadly generalizable, depending only on user-defined intermolecular connectivity, is operable across different length scales, and is extensible to the study of dynamic phenomena.

Multiscale Very High Resolution Topographic Models in Alpine Ecology: Pros and Cons of Airborne LiDAR and Drone-Based Stereo-Photogrammetry Technologies

The vulnerability of alpine environments to climate change presses an urgent need to accurately model and understand these ecosystems. Popularity in the use of digital elevation models (DEMs) to derive proxy environmental variables has increased over the past decade, particularly as DEMs are relatively cheaply acquired at very high resolutions (VHR; <1 m spatial resolution). Here, we implement a multiscale framework and compare DEM-derived variables produced by Light Detection and Ranging (LiDAR) and stereo-photogrammetry (PHOTO) methods, with the aim of assessing their relevance and utility in species distribution modelling (SDM). Using a case study on the arctic-alpine plant, Arabis alpina, in two valleys in the western Swiss Alps, we show that both LiDAR and PHOTO technologies can be relevant for producing DEM-derived variables for use in SDMs. We demonstrate that PHOTO DEMs, up to a spatial resolution of at least 1 m, rivalled the accuracy of LiDAR DEMs, largely owing to the customizability of PHOTO DEMs to the study sites compared to commercially available LiDAR DEMs. We obtained DEMs at spatial resolutions of 6.25 cm–8 m for PHOTO and 50 cm–32 m for LiDAR, where we determined that the optimal spatial resolutions of DEM-derived variables in SDM were between 1 and 32 m, depending on the variable and site characteristics. We found that the reduced extent of PHOTO DEMs altered the calculations of all derived variables, which had particular consequences on their relevance at the site with heterogenous terrain. However, for the homogenous site, SDMs based on PHOTO-derived variables generally had higher predictive powers than those derived from LiDAR at matching resolutions. From our results, we recommend carefully considering the required DEM extent to produce relevant derived variables. We also advocate implementing a multiscale framework to appropriately assess the ecological relevance of derived variables, where we caution against the use of VHR-DEMs finer than 50 cm in such studies.

Potential impact assessment between sst and extreme precipitation: an observational study in Southeastern Brazil

This work aims to identify, by means of statistical analysis, the relationship between extreme precipitation events and SST anomalies (SSTA) in the municipalities surrounding the Ilha Grande (IGB) and Sepetiba (SB) bays in the state of Rio de Janeiro evaluating its annual distributions using measured data from rainfall stations and relating these events to potential SST anomalies found in this region. Data from 10 rainfall stations distributed around the bays and SST dataset obtained from the Multi-scale Ultra-high Resolution Sea Surface Temperature (MUR) from 2003 to 2014 were used. SSTA and their average were generated for each of the three subareas delimited in the study area. The obtained time series of each subarea define the thresholds for significant SSTA, based on their respective standard deviation. The number of days with SSTA were obtained from these significant anomaly values, and days with positive or negative SSTA, for each subarea, were computed. For SSTA analysis, only events with significant anomalies were considered. Only SSTA occurring on the days of intense precipitation highlighted in P95 and up to 3 days before the extreme event date were analyzed. The results had shown that about 35% of extreme precipitation events were potentially related to the presence of significant SST anomalies in the studied area. Finally, it is pointed that regional anomalies in SST, such as those described in this study, can also contribute to the formation of extreme rain events.

Impacts of the Kuroshio Intrusion through the Luzon Strait on the Local Precipitation Anomaly

The Kuroshio Current has its origin in the northwestern Pacific, flowing northward to the east of Taiwan and the northern part of Luzon Island. As the Kuroshio Current flows northward, it quasi-periodically intrudes (hereafter referred to as Kuroshio intrusion (KI)) into the northern South China Sea (SCS) basin through the Luzon Strait. Despite the complex generation mechanisms of KI, the purpose of this study is to improve our understanding of the effects of KI through the Luzon Strait on the regional atmospheric and weather variations. Long-term multiple satellite observations, including absolute dynamic topography, absolute geostrophic currents, sea surface winds by ASCAT, multi-scale ultra-high resolution sea surface temperature (MURSST) level-four analysis, and research-quality three-hourly TRMM multi-satellite precipitation analysis (TMPA), was used to systematically examine the aforementioned scientific problem. Analysis indicates that the KI is interlinked with the consequential anomalous precipitation off southwestern Taiwan. This anomalous precipitation would lead to ~560 million tons of freshwater influx during each KI event. Subsequently, independent moisture budget analysis suggests that moisture, mainly from vertical advection, is the possible source of the precipitation anomaly. Additionally, a bulk formula analysis was applied to understand how KI can trigger the precipitation anomaly through vertical advection of moisture without causing an evident change in the low-level flows. These new research findings might reconcile the divisiveness on why winds are not showing a synchronous response during the KI and consequential anomalous precipitation events.

Improvement of the WENO-Z+ scheme

The WENO-Z+ scheme (J. Comput. Phys. 2016, 313: 726-753) was obtained by adding a new term into the WENO-Z weights. The added term has the role of raising the weights of less-smooth substencils. The WENO-Z+ scheme achieves superior results in the test problems that involve only a single wave component. For problems containing multiscale structures, however, the results of WENO-Z+ show little improvement compared with those of WENO-Z. In this short note, we investigate the reason why WENO-Z+ has little improvement in solving the multiscale problems and propose a set of modifications to it. A series of numerical tests confirm that the new scheme, which is named WENO-Z+I, has significantly improved multiscale resolution. The multiscale resolution of the fifth-order WENO-Z+I scheme is even significantly better than that of the seventh-order WENO-Z scheme. We also find that the WENO-Z+ type schemes can achieve better spectral properties than the corresponding linear upwind scheme.

Seasonal Variability of SST Fronts in the Inner Sea of Chiloé and Its Adjacent Coastal Ocean, Northern Patagonia

Surface oceanic fronts are regions characterized by high biological activity. Here, Sea Surface Temperature (SST) fronts are analyzed for the period 2003–2019 using the Multi-scale Ultra-high Resolution (MUR) SST product in northern Patagonia, a coastal region with high environmental variability through river discharges and coastal upwelling events. SST gradient magnitudes were maximum off Chiloé Island in summer and fall, coherent with the highest frontal probability in the coastal oceanic area, which would correspond to the formation of a coastal upwelling front in the meridional direction. Increased gradient magnitudes in the Inner Sea of Chiloé (ISC) were found primarily in spring and summer. The frontal probability analysis revealed the highest occurrences were confined to the northern area (north of Desertores Islands) and around the southern border of Boca del Guafo. An Empirical Orthogonal Function analysis was performed to clarify the dominant modes of variability in SST gradient magnitudes. The meridional coastal fronts explained the dominant mode (78% of the variance) off Chiloé Island, which dominates in summer, whereas the SST fronts inside the ISC (second mode; 15.8%) were found to dominate in spring and early summer (October–January). Future efforts are suggested focusing on high frontal probability areas to study the vertical structure and variability of the coastal fronts in the ISC and its adjacent coastal ocean.

Sea Ice Properties in High-Resolution Sea Ice Models.

An Arctic sea ice-ocean model is run with three uniform horizontal resolutions (6, 4, and 2km) and identical sea ice and ocean model parameterizations, including an isotropic viscous-plastic sea ice rheology, a mechanical ice strength parameterization, and an ice ridging parameterization. Driven by the same atmospheric forcing, the three model versions all produce similar spatial patterns and temporal variations of ice thickness and motion fields, resulting in almost identical magnitude and seasonal evolution of total ice volume and mean ice concentration, ice speed, and fractions of ice of various thickness categories over the Arctic Ocean. Increasing model resolution from 6 to 2km does not significantly improve model performance when compared to NASA IceBridge ice thickness observations. This suggests that the large-scale sea ice properties of the model are insensitive to varying high resolutions within the multifloe scale (2-10km), and it may be unnecessary to adjust model parameters constantly with increasingly high resolutions. This is also true with models within the aggregate scale (10-75km), indicating that model parameters used at coarse resolution may be used at high or multiscale resolution. However, even though the three versions all yield similar mean state of sea ice, they differ in representing anisotropic properties of sea ice. While they produce a basic pattern of major sea ice leads similar to satellite observations, their leads are distributed differently in space and time. Without changing model parameters and sea ice spatiotemporal variability, the 2-km resolution model tends to capture more leads than the other two models.

FTNet: Feature Transverse Network for Thermal Image Semantic Segmentation

Thermal imaging is a process of using infrared radiation and thermal energy to collect information about objects. It is superior to visible imaging for its ability to operate in darkness and tolerate illumination variations. In addition, it has potential to penetrate smoke, aerosol, dust, and mist, which are critical inhibitors for visible imaging applications, including semantic segmentation. Unfortunately, current state-of-the-art image semantic segmentation methods (i) mainly concentrate on visible spectrum images, which do not adequately capture the context of corresponding pixels, particularly edge details in thermal images, and (ii) accept a trade-off between higher accuracy and lower speed, or vice-versa. Here, a novel end-to-end trainable convolutional neural network architecture, feature transverse network (FTNet), has been proposed to solve the aforementioned problems. FTNet captures and optimizes feature representation at the multi-scale resolution, thereby improving the capability to process high-resolution images and producing quality output with a lower computational cost. Extensive computer experimentations were conducted on publicly available benchmarking thermal datasets, including SODA, MFNet, and SCUT-Seg, to demonstrate the effectiveness of the proposed FTNet compared to state-of-the-art methods. This comparison includes multiple aspects, including the quantitative accuracy and speed of the various approaches. The source code is available at https://github.com/shreyaskamathkm/FTNet .

Automated Skin Lesion Segmentation Via an Adaptive Dual Attention Module.

We present a convolutional neural network (CNN) equipped with a novel and efficient adaptive dual attention module (ADAM) for automated skin lesion segmentation from dermoscopic images, which is an essential yet challenging step for the development of a computer-assisted skin disease diagnosis system. The proposed ADAM has three compelling characteristics. First, we integrate two global context modeling mechanisms into the ADAM, one aiming at capturing the boundary continuity of skin lesion by global average pooling while the other dealing with the shape irregularity by pixel-wise correlation. In this regard, our network, thanks to the proposed ADAM, is capable of extracting more comprehensive and discriminative features for recognizing the boundary of skin lesions. Second, the proposed ADAM supports multi-scale resolution fusion, and hence can capture multi-scale features to further improve the segmentation accuracy. Third, as we harness a spatial information weighting method in the proposed network, our method can reduce a lot of redundancies compared with traditional CNNs. The proposed network is implemented based on a dual encoder architecture, which is able to enlarge the receptive field without greatly increasing the network parameters. In addition, we assign different dilation rates to different ADAMs so that it can adaptively capture distinguishing features according to the size of a lesion. We extensively evaluate the proposed method on both ISBI2017 and ISIC2018 datasets and the experimental results demonstrate that, without using network ensemble schemes, our method is capable of achieving better segmentation performance than state-of-the-art deep learning models, particularly those equipped with attention mechanisms.