Deep Shadow Research Articles

Deep unsupervised shadow detection with curriculum learning and self-training

Shadow detection is undergoing a rapid and remarkable development along with the wide use of deep neural networks. Benefiting from a large number of training images annotated with strong pixel-level ground-truth masks, current deep shadow detectors have achieved state-of-the-art performance. However, it is expensive and time-consuming to provide the pixel-level ground-truth mask for each training image. Considering that, this paper proposes the first unsupervised deep shadow detection framework, which consists of an initial pseudo label generation (IPG) module, a curriculum learning (CL) module and a self-training (ST) module. The supervision signals used in our learning framework are generated from several existing traditional unsupervised shadow detectors, which usually contain a lot of noisy information. Therefore, each module in our unsupervised framework is dedicated to reduce the adverse influence of noisy information on model training. Specifically, the IPG module combines different traditional unsupervised shadow maps to obtain their complementary shadow information. After obtaining the initial pseudo labels, the CL module and the ST module will be used in conjunction to gradually learn new shadow patterns and update the qualities of pseudo labels simultaneously. Extensive experimental results on various benchmark datasets demonstrate that our deep shadow detector not only outperforms the traditional unsupervised shadow detection methods by a large margin but also achieves comparable results with some recent state-of-the-art fully-supervised deep shadow detection methods.

Synthetic Document Images with Diverse Shadows for Deep Shadow Removal Networks.

Shadow removal for document images is an essential task for digitized document applications. Recent shadow removal models have been trained on pairs of shadow images and shadow-free images. However, obtaining a large, diverse dataset for document shadow removal takes time and effort. Thus, only small real datasets are available. Graphic renderers have been used to synthesize shadows to create relatively large datasets. However, the limited number of unique documents and the limited lighting environments adversely affect the network performance. This paper presents a large-scale, diverse dataset called the Synthetic Document with Diverse Shadows (SynDocDS) dataset. The SynDocDS comprises rendered images with diverse shadows augmented by a physics-based illumination model, which can be utilized to obtain a more robust and high-performance deep shadow removal network. In this paper, we further propose a Dual Shadow Fusion Network (DSFN). Unlike natural images, document images often have constant background colors requiring a high understanding of global color features for training a deep shadow removal network. The DSFN has a high global color comprehension and understanding of shadow regions and merges shadow attentions and features efficiently. We conduct experiments on three publicly available datasets, the OSR, Kligler's, and Jung's datasets, to validate our proposed method's effectiveness. In comparison to training on existing synthetic datasets, our model training on the SynDocDS dataset achieves an enhancement in the PSNR and SSIM, increasing them from 23.00 dB to 25.70 dB and 0.959 to 0.971 on average. In addition, the experiments demonstrated that our DSFN clearly outperformed other networks across multiple metrics, including the PSNR, the SSIM, and its impact on OCR performance.

A Review of Advancements in Deep Learning-Based Shadow Detection and Removal in Image and Video Analysis

A Review of Advancements in Deep Learning-Based Shadow Detection and Removal in Image and Video Analysis

A review of advancements in deep learning-based shadow detection and removal in image and video analysis

A review of advancements in deep learning-based shadow detection and removal in image and video analysis

Retracted: Deep Learning-Based Multitarget Motion Shadow Rejection and Accurate Tracking for Sports Video

Retracted: Deep Learning-Based Multitarget Motion Shadow Rejection and Accurate Tracking for Sports Video

ShadowFormer: Global Context Helps Shadow Removal

Recent deep learning methods have achieved promising results in image shadow removal. However, most of the existing approaches focus on working locally within shadow and non-shadow regions, resulting in severe artifacts around the shadow boundaries as well as inconsistent illumination between shadow and non-shadow regions. It is still challenging for the deep shadow removal model to exploit the global contextual correlation between shadow and non-shadow regions. In this work, we first propose a Retinex-based shadow model, from which we derive a novel transformer-based network, dubbed ShandowFormer, to exploit non-shadow regions to help shadow region restoration. A multi-scale channel attention framework is employed to hierarchically capture the global information. Based on that, we propose a Shadow-Interaction Module (SIM) with Shadow-Interaction Attention (SIA) in the bottleneck stage to effectively model the context correlation between shadow and non-shadow regions. We conduct extensive experiments on three popular public datasets, including ISTD, ISTD+, and SRD, to evaluate the proposed method. Our method achieves state-of-the-art performance by using up to 150X fewer model parameters.

Underwater source localisation utilising interference pattern under low SNR conditions

AbstractUnderwater acoustic source localisation in deep water shadow zone is difficult since the vertical aperture of the array is limited and the received signal‐noise ratio (SNR) is low. This article investigates a novel interference pattern in the first shadow zone. The intensity displays two oscillating striations in the interference pattern, which can be used to estimate the source depth under the restriction of lacking adequate spatial samples and acoustic environmental parameters, and the range of the source can be estimated roughly using the position of which. The method is applicable to multi‐target localisation. In practice, the interference pattern is always corrupted by noise under low SNR conditions, resulting in performance degradation or even failure of the localisation. To address the problem of recovering the corrupted interference pattern, an interference tensor is constructed and analysed, which expands the general two‐dimensional interference pattern into range‐grazing angle‐frequency domain to make use of the multi‐frame interaction coherently. A Tucker decomposition‐based method that adopts a low‐rank representation of the interference tensor is proposed to retrieve the intact interference pattern. Simulated data validate the accuracy of the localisation under low SNR conditions. It is shown that this method significantly improves the estimated performance over conventional matched‐field processing.

SEAT-YOLO: A squeeze-excite and spatial attentive you only look once architecture for shadow detection

Recently, several deep learning-based architectures have been utilized for the recognition of shadow regions in images and videos with the aim of classification and segmentation of shadow instances. The present work aims towards solving the problem of shadow recognition as a detection and regression problem. To solve the problem of shadow detection in images and videos, we have developed SEAT-YOLO, a deep learning-based shadow detection architecture. The backbone network of the proposed architecture is based on convolution layers, squeeze-excite blocks, spatial attention module, and spatial pyramid pooling layer and the detection network utilizes YOLO detection heads. The proposed SEAT-YOLO architecture is trained and tested on the publicly available SBU shadow detection dataset. For shadow detection on the SBU dataset, it achieved a mAP value of 59.53 %. Furthermore, the proposed SEAT-YOLO is capable of detecting small shadow regions with high accuracy which is a challenge with most deep learning-based detection architectures. Moreover, the other advantage of the proposed SEAT-YOLO is its ability to detect multiple shadow regions in a single image by drawing precise bounding boxes for overlapping shadows. The proposed SEAT-YOLO architecture has high practical implications in driverless vehicles for detection of shadow regions and take effective driving decisions.

Structure-Informed Shadow Removal Networks.

Existing deep learning-based shadow removal methods still produce images with shadow remnants. These shadow remnants typically exist in homogeneous regions with low-intensity values, making them untraceable in the existing image-to-image mapping paradigm. We observe that shadows mainly degrade images at the image-structure level (in which humans perceive object shapes and continuous colors). Hence, in this paper, we propose to remove shadows at the image structure level. Based on this idea, we propose a novel structure-informed shadow removal network (StructNet) to leverage the image-structure information to address the shadow remnant problem. Specifically, StructNet first reconstructs the structure information of the input image without shadows and then uses the restored shadow-free structure prior to guiding the image-level shadow removal. StructNet contains two main novel modules: (1) a mask-guided shadow-free extraction (MSFE) module to extract image structural features in a non-shadow-to-shadow directional manner, and (2) a multi-scale feature & residual aggregation (MFRA) module to leverage the shadow-free structure information to regularize feature consistency. In addition, we also propose to extend StructNet to exploit multi-level structure information (MStructNet), to further boost the shadow removal performance with minimum computational overheads. Extensive experiments on three shadow removal benchmarks demonstrate that our method outperforms existing shadow removal methods, and our StructNet can be integrated with existing methods to improve them further.

Introduction

The queen’s court was modeled on the king’s court and, as Marcin Kromer wrote in the 16th century, the ruler “feeds, clothes and remunerates” of all his wife’s courtiers (Kromer, 1984, 147), due to its role to merely serve its mistress, it could not be similar to the king’s court in terms of its role in the state. It seems that due to this disparity, the actual role of the queen’s court in politics, culture or the economy was mostly determined by her personality and ambitions on top of her relationship with her spouse, which increased or decreased her chances of making her court, or more precisely, the people associated with it, an important political factor. It is not surprising, therefore, that the attention of researchers was attracted to those Polish queens who stood out for their activity and influence on the state, for example, such as Queen Bona Sforza, while those who rarely or never went beyond the role of a royal consort remained in a deep shadow. Although it should be noted at once that there was an obvious change when the queen’s spouse also became the queen mother – the need to secure the fate of her children, especially sons, generally motivated Polish queens and those immediately surrounding them, making their courts and the people associated with them active participants in public life (SucheniGrabowska, 2009, pp. 247–260). However, these issues did not receive separate attention.

Marquis de Sade, The Marquise de Gange. Translated with an Introduction and notes by Will McMorran

This edition retrieves Sade’s late novel from the deep shadow of his pornographic production and places it squarely within the gothic canon. In so doing, it pursues a line also productively explored by Philippe Roger in his essay on Sade and the gothic in John D. Lyons’s collection The Dark Thread: From Tragical Histories to Gothic Tales (Newark: ­University of Delaware Press, 2019; reviewed in FS, 75 (2021), 405–06). Will McMorran’s excellent Introduction dramatically retells the story of the notorious true crime in which a Provençal noblewoman was murdered by her husband’s brothers in 1667, showing where Sade’s tale fits within the many accounts of her death. He also provides a very helpful selected bibliography of the retellings of the story, of which there are many from both before and after Sade, most recently in 1984, supplying in its fascinating entirety an English translation from 1744 of François Gayot de Pitaval’s influential version. ­Notable (re-)writers include Charlotte Smith, Alexandre Dumas père, and Elizabeth Gaskell. ­McMorran’s argument is two-pronged: firstly, that this novel is not of lesser importance to Sadean studies than the earlier works, and here he joins other recent scholars looking to rehabilitate the late novels (notably Michèle Vallenthini’s Sade dans l’histoire: du temps de la fiction à la fiction du temps (Paris: Classiques Garnier, 2019)). His second goal is to establish Sade as a properly gothic author, influenced both by the English gothic and by what might be seen as the genre’s origins in the Histoires tragiques tradition. And the result, in McMorran’s view, is that with The Marquise de Gange ‘the Gothic is coming home’ (p. xxv). So, what of the novel itself? Castles, dank rooms, fell premonitions, and a beautiful heroine unfairly treated are all abundantly present. The reader familiar with Sade will also recognize the disquisitions on nature and won’t be too surprised to discover that the ‘dissolute’ man is ‘inflamed rather than soothed’ by the beautiful ‘apologist for virtue’ (p. 90). There are some bravura passages of gothic horror and in one such we find the Marquise espying a hidden door: ‘a spluttering lamp reveals a glimpse of a chamber on the other side of the door; she enters… But what hideous object greets her eyes! There on a table before her, she sees an open cadaver, almost entirely ripped asunder’ (p. 130, punctuation sic). She escapes for the time being, yet ‘tottering, buffeted both by the gale and her own terror, [she] resembles a young willow battered by a storm’ (p. 131). Poor woman! Her horrid fate will not be eluded. In amongst all this turbulence, the (­relatively) sober account (p. 80) of what it’s like to be imprisoned adds depth and feeling, and if I can’t agree with McMorran that The Marquise de Gange offers a ‘taut, tense narrative’ (p. xxvii), he is right that this is a text worth revisiting and which should be part of a fuller sense of the gothic canon, on both sides of the English Channel.

Our Lady of Innumerable Appellatives

1.Swirl of gold gleaming in our daughters’ eyes.Amen. Whorl of cirrostratus haloingthe moon. Amen. Your fingerprints areeverywhere. Lady of Ice. Lady of Sand.of Maple Star and Gingko Fan. of Rot.of Scat. of Hoarders and Whores. Goddess,is there any title you would refuse?of Shopping Carts with One Wonky Wheel.of Loofas. of Wedding Cakes. of ArtisanalPumpernickel Loaves. of Mud. of Moss.of Mirages. of Vultures and of Vespers.We'd like to pour you into a ceramic mugwe made in high school, sloppy rings and drippy glaze.We'd like to tuck you into a locket,2.fold you tiny into a locket,wear you around our necks, the ultimateamulet. You're no saint of narrow scope.Your sight is wide, your praxis: gatherhood.of Violin recitals and the A String Concerto.of Messy Music and Divine Harmony.of Cacophony. of Canticle. of Beginner's luck.Dead finch on the sidewalk under a bloomingmagnolia tree. of Ovation andOration and Oblation and Elevationand Desecration. Carpenter ants.Water bears. Leptons. Lady of SmallestUnits and Magnificatus Extremis.Lady of Made Up Latin. of Lingua Franca.3.Teach us your lingua franca, your pidgin.Lady of Tamil. Lady of Greek.of Nahuatl. of Sanskrit. of Urdu.Eskimo. Korean. Hebrew.Chinese. Egyptian. Nushu.Lady of Mariology. Angelology.Eschatology. All the -ologiesand all the -olatries. Lunolatryand Icthyolatry. Astrolatryand Zoolatry. Lady of Words. TheWord. of Celery, even, the fibrous kindimpossible to chew. of Math, ComplexCalculus, mind-bending Theoretical Physics.All that, plus simpler stuff, like addition.4.Lady of the Simple Truth of Addition,like loneliness +1 and what that equals.Lady of Every Possible Love.Same sex. Trans. Bi. Asexual. Pansexual.Lady of All Loving Formulations &Goddess of Every Gender. Transgender.Cisgender. Agender. Pangender. Non-binary. Two-spirit. Genderqueer. All the genders.Holiness of every color. Black. Pink. Red. Blue. Grey.All the colors and color combinations.All the ages, zygote to crone. Goddessof Puberty, of Menstruation, of Menopause.Goddess of the Blood Moon, bleed us, lead usthrough the shadow of your pulsing seasons.5.Through deep shadow of each pulsing season,lead us, Lady of Every Kind of Blood.A. B. O. Positive. Negative. Anemic.Cold-blooded. Warm-blooded. Sap-blooded.Overstory and understory.Root, trunk, branch, foliage. Sky and earth.Foot, torso, arm, head. Death and birth. Goddessof Endless Litanies, we like to imagineyour special grace lighting our darkestcorners, the forgotten cobwebbed crevices.We imagine your bright gaze burning updew on the neglected rake. We hold tightto the myth of specialness, chosendom.But, Madonna, you're no leashed deity.6.Lady, you're no leashed deity. No chainforged between you and our basin. Our grove.Our desert. Our field. Our mountain. Free-rangeGoddess, you've answered to many names.Asherah. Innana. Ishtar. Astarte.Sauska. Isis. Venus. Ema. Sophia. Shekinah.Semiramis. Amaterasu. Aphrodite.Queen of the Night. Queen of Heaven. A hostof others we've lost or forgotten. Nowwe cry out with tears Lady, I believe;help thou my unbelief. Swallow uslike tepid water. Hold us in your handslike tufts of fog. Tuck your song into ourwooden bodies, you the string, you the bow.Teach us music fluid enough for praise.7.Teach us music fluid enough to paintyour praise. Gift us eyes wide enough to drinkyour crystalline wonders. Stretch our earstuba-huge to catch your spirit-whispers.Make of our ragged patches one quilt warmenough for every dreamer: rock, bug, sprout.You are the Shepherdess knocking iciclesfrom our wooly locks. We are the sheep,sure of sheer, oblivious to your care,Our Lady of Ice. Our Lady of Sand.of Maple Star and Gingko Fan.of. of. of. Your fingerprints everywhere:Cloud-whorl, a halo round the moon. Amen.Gold-swirl, our daughters’ radiant eyes. Amen.

C2ShadowGAN: cycle-in-cycle generative adversarial network for shadow removal using unpaired data

Recent advances in deep learning technology, and the availability of public shadow image datasets, have enabled significant performance improvements of shadow removal tasks in computer vision. However, most deep learning-based shadow removal methods are usually trained in a supervised manner, in which paired shadow and shadow-free data are required. We developed a weakly supervised generative adversarial network with a cycle-in-cycle structure for shadow removal using unpaired data. In addition, we introduced new loss functions to reduce unnecessary transformations for non-shadow areas and to enable smooth transformations for shadow boundary areas. We conducted extensive experiments using the ISTD and Video Shadow Removal datasets to assess the effectiveness of our methods. The experimental results show that our method is superior to other state-of-the-art methods trained on unpaired data.

Choosing Brave: How Mamie Till-Mobley and Emmett Till Sparked the Civil Rights Movement by Angela Joy

Reviewed by: Choosing Brave: How Mamie Till-Mobley and Emmett Till Sparked the Civil Rights Movement by Angela Joy Kate Quealy-Gainer, Editor Joy, Angela Choosing Brave: How Mamie Till-Mobley and Emmett Till Sparked the Civil Rights Movement; illus. by Janelle Washington. Roaring Brook, 2022 [64p] Trade ed. ISBN 9781250220950 $19.99 E-book ed. ISBN 9781250893673 $10.99 Reviewed from digital galleys R* Gr. 3-5 The epigraph of this poignant biography—"I just had to pray and brave it"—embodies and is attributed to its subject, Mamie Till-Mobley, a woman who chose to make her personal pain politically relevant. It's a unique focus, stepping away from the horrific details that made Emmett Till's death such a lightning rod for the civil rights movement to instead examine his mother's life before and after the loss of her son. Intelligent and determined, Mamie nonetheless dealt with her fair share of hardships, including the abandonment of her father and the abuse by and then death of her husband. But it's clear Emmett was her pride and joy; readers see Mamie encourage his rambunctiousness, nurture him through illness—and teach him to whistle through his stutter, a trick that might have ultimately attracted the attention of his killers. She translated her grief into power: "it would have been easy, then, to go to sleep beside him; to save her pain for privacy … But Mamie did the braver thing. She sounded the alarm." The book is careful to admire that bravery and endurance but quick to grieve the fact that she needed to muster such strength at all. By placing Emmett's murder in the context of her life, the narrative brings a deep intimacy to the portrayal of a woman mourning her son, not a symbol. The cut paper art is startling in its use of negative space, casting the scene in which Mamie learns of Emmett's death in deep shadow and later pulling in layers of red as the jury's acquittal of his murderers is announced. The last illustrated page, featuring the silhouette of a woman at church surrounded by the names of Black people killed recently in hate crimes, is made more gut-wrenching by the author and illustrator notes. [End Page 52] Copyright © 2022 The Board of Trustees of the University of Illinois

Ray Optical Scattering From Uniform Reflective Cylindrical Metasurfaces Using Surface Susceptibility Tensors

A ray optical methodology based on the uniform theory of diffraction (UTD) is proposed to model electromagnetic (EM) field scattering from curved metasurfaces (MSs). The problem addressed is the illumination of a purely reflective uniform cylindrical MS by a line source, models the surface with susceptibilities and employs a methodology previously used for cylinders coated in thin dielectric layers [Kim and Wang (1989)]. The approach is fundamentally based on a representation of the MS using the generalized sheet transition conditions (GSTCs) which characterizes the surface in terms of susceptibility dyadics. An eigenfunction (EF) description of the MS problem is derived considering both tangential and normal surface susceptibilities, and used to develop a ray optics (RO) description of the scattered fields including the specular geometrical optical field, surface diffraction described by creeping waves and a transition region over the shadow boundary. The specification of the fields in the transition region is dependent on the evaluation of the Pekeris caret function integral and the method follows [Kim and Wang (1989)]. The proposed RO-GSTC model is then successfully demonstrated for a variety of cases and is independently verified using a rigorous EF solution (EF-GSTC) and full-wave Integral Equation method (IE-GSTC), over the entire domain from the deep lit to deep shadow.

Evaluating Statewide NAIP Photogrammetric Point Clouds for Operational Improvement of National Forest Inventory Estimates in Mixed Hardwood Forests of the Southeastern U.S.

The U.S. Forest Service, Forest Inventory and Analysis (FIA) program is tasked with making and reporting estimates of various forest attributes using a design-based network of permanent sampling plots. To make its estimates more precise, FIA uses a technique known as post-stratification to group plots into more homogenous classes, which helps lower variance when deriving population means. Currently FIA uses a nationally available map of tree canopy cover for post-stratification, which tends to work well for forest area estimates but less so for structural attributes like volume. Here we explore the use of new statewide digital aerial photogrammetric (DAP) point clouds developed from stereo imagery collected by the National Agricultural Imagery Program (NAIP) to improve these estimates in the southeastern mixed hardwood forests of Tennessee and Virginia, United States (U.S.). Our objectives are to 1. evaluate the relative quality of NAIP DAP point clouds using airborne LiDAR and FIA tree height measurements, and 2. assess the ability of NAIP digital height models (DHMs) to improve operational forest inventory estimates above the gains already achieved from FIA’s current post-stratification approach. Our results show the NAIP point clouds were moderately to strongly correlated with FIA field measured maximum tree heights (average Pearson’s r = 0.74) with a slight negative bias (−1.56 m) and an RMSE error of ~4.0 m. The NAIP point cloud heights were also more accurate for softwoods (R2s = 0.60–0.79) than hardwoods (R2s = 0.33–0.50) with an error structure that was consistent across multiple years of FIA measurements. Several factors served to degrade the relationship between the NAIP point clouds and FIA data, including a lack of 3D points in areas of advanced hardwood senescence, spurious height values in deep shadows and imprecision of FIA plot locations (which were estimated to be off the true locations by +/− 8 m). Using NAIP strata maps for post-stratification yielded forest volume estimates that were 31% more precise on average than estimates stratified with tree canopy cover data. Combining NAIP DHMs with forest type information from national map products helped improve stratification performance, especially for softwoods. The monetary value of using NAIP height maps to post-stratify FIA survey unit total volume estimates was USD 1.8 million vs. the costs of installing more field plots to achieve similar precision gains. Overall, our results show the benefit and growing feasibility of using NAIP point clouds to improve FIA’s operational forest inventory estimates.

Rethinking Academic Conferences in the Age of Pandemic

The year 2020 witnessed the havoc wreaked by the coronavirus disease COVID-19 due to its onset in late 2019. The COVID-19 pandemic is the cruelest public health crisis humankind has ever seen. The COVID-19 pandemic profoundly affected every walk of life, and academic research has been no exception. Academic conferences are an indispensable component of research. Note that the pandemic together with its variants ravaged the globe in 2020, while their recurrences yet have a deep shadow across 2021 and 2022 with uncertainties for the near future. Under the sway of the pandemic, many conferences are conducted in virtual mode to mitigate the propagation of the virus. It is no surprise that academic conferences charge the attendees for registration fees with the amount varying by countries and disciplines. Here, we collect the registration fee information for conferences held in 2019, 2020 and 2021. Note that virtual conferences barely cater to attendees except by providing online platforms. However, we discover that most of the virtual conferences held in 2020 and 2021 still charged high registration fees compared to those in 2019, while the remaining conferences only applied small discounts. In light of the current situation of the pandemic as well as uncertainties in the future, virtual conferences could be a common form of academic activity. Considering the sluggish global economy at well as other potential issues, here, we advocate that going virtual should always be an option for academic conferences in the future. We also suggest that virtual conferences should charge less and the expenditure of the fees should be open to the public.

The Private Investigator

The Private Investigator

Britain’s ‘Dark Factories’: Specters of Racial Capitalism Today

Recent revelations about illegally low wages and abject working conditions in Leicester’s “dark” garment factories—endured by its vulnerable mainly global majority and migrant workforce—have once again highlighted British fashion’s benighted “other” and its reliance on cheapened and “expendable” “racialized” sweated labor. Forced to work on through the Covid pandemic to produce fast fashion for e-commerce brands, “British” fashion manufacture now reflects the wider plight of the global CMT (Cut, Make and Trim) army, who at considerable personal risk and cost, manufacture most of our fashion. Drawing on a set of theoretical innovations in Marxist and post-Marxist theory, notably Smith and Suwandi theory of labor “super-exploitation,” De Genova’s work on the disciplining of migrants through “border regimes and spectacles” and Gargi Bhattacharyya’s critical reading of Cedric Robinson’s thesis of “Racial Capitalism” (1983), this article explains how and why its specter still castes such a deep and troubling shadow over the production of fashion.

Teaching Fairy Tales

Teaching Fairy Tales