Object Discrimination Research Articles

Sensorless Object Discrimination by a Particle Filter for Caging-based Part Feeders

Sensorless Object Discrimination by a Particle Filter for Caging-based Part Feeders

Pixel-Level Domain Adaptation: A New Perspective for Enhancing Weakly Supervised Semantic Segmentation.

Recent attention has been devoted to the pursuit of learning semantic segmentation models exclusively from image tags, a paradigm known as image-level Weakly Supervised Semantic Segmentation (WSSS). Existing attempts adopt the Class Activation Maps (CAMs) as priors to mine object regions yet observe the imbalanced activation issue, where only the most discriminative object parts are located. In this paper, we argue that the distribution discrepancy between the discriminative and the non-discriminative parts of objects prevents the model from producing complete and precise pseudo masks as ground truths. For this purpose, we propose a Pixel-Level Domain Adaptation (PLDA) method to encourage the model in learning pixel-wise domain-invariant features. Specifically, a multi-head domain classifier trained adversarially with the feature extraction is introduced to promote the emergence of pixel features that are invariant with respect to the shift between the source (i.e., the discriminative object parts) and the target (i.e., the non-discriminative object parts) domains. In addition, we come up with a Confident Pseudo-Supervision strategy to guarantee the discriminative ability of each pixel for the segmentation task, which serves as a complement to the intra-image domain adversarial training. Our method is conceptually simple, intuitive and can be easily integrated into existing WSSS methods. Taking several strong baseline models as instances, we experimentally demonstrate the effectiveness of our approach under a wide range of settings.

Learning to Aggregate Multi-Scale Context for Instance Segmentation in Remote Sensing Images.

The task of instance segmentation in remote sensing images, aiming at performing per-pixel labeling of objects at the instance level, is of great importance for various civil applications. Despite previous successes, most existing instance segmentation methods designed for natural images encounter sharp performance degradations when they are directly applied to top-view remote sensing images. Through careful analysis, we observe that the challenges mainly come from the lack of discriminative object features due to severe scale variations, low contrasts, and clustered distributions. In order to address these problems, a novel context aggregation network (CATNet) is proposed to improve the feature extraction process. The proposed model exploits three lightweight plug-and-play modules, namely, dense feature pyramid network (DenseFPN), spatial context pyramid (SCP), and hierarchical region of interest extractor (HRoIE), to aggregate global visual context at feature, spatial, and instance domains, respectively. DenseFPN is a multi-scale feature propagation module that establishes more flexible information flows by adopting interlevel residual connections, cross-level dense connections, and feature reweighting strategy. Leveraging the attention mechanism, SCP further augments the features by aggregating global spatial context into local regions. For each instance, HRoIE adaptively generates RoI features for different downstream tasks. Extensive evaluations of the proposed scheme on iSAID, DIOR, NWPU VHR-10, and HRSID datasets demonstrate that the proposed approach outperforms state-of-the-arts under similar computational costs. Source code and pretrained models are available at https://github.com/yeliudev/CATNet.

Boosting Weakly Supervised Object Localization and Segmentation With Domain Adaption.

Weakly supervised object localization (WSOL), adopting only image-level annotations to learn the pixel-level localization model, can release human resources in the annotation process. Most one-stage WSOL methods learn the localization model with multi-instance learning, making them only activate discriminative object parts rather than the whole object. In our work, we attribute this problem to the domain shift between the training and test process of WSOL and provide a novel perspective that views WSOL as a domain adaption (DA) task. Under this perspective, a DA-WSOL pipeline is elaborated to better assist WSOL with DA approaches by considering the specificities for the adaption of WSOL. Our DA-WSOL pipeline can discern the source-related and the Universum samples from other target samples based on a proposed target sampling strategy and then utilize them to solve the sample unbalancing and label unmatching between the source and target domain of WSOL. Experiments show that our pipeline outperforms SOTA methods on three WSOL benchmarks and can improve the performance of downstream weakly supervised semantic segmentation tasks. Codes are available at https://github.com/zh460045050/dawsol.

Quantitative Phytochemistry and Neuro- histological Effect of Tamarindus indica Fruit Pulp Aqueous Extract on Traumatic Brain Injury in Albino Rats

Tamarindus indica has numerous therapeutic benefits. Its therapeutic effects have been reported in some neurodegenerative conditions like Alzheimer's disease in rats. This study evaluated quantitative phytochemistry and effect of tamarind fruit pulp extract in traumatic brain injury (TBI) - induced neurobehavioral and histological changes in rats. Six groups of five rats each were used for this study. Groups I, II, III, and IV were treated with tamarind fruit pulp extract at doses of 100, 200, 400, and 800 mg/kg, orally, after being induced with TBI. Rats in group V were traumatized but not treated (TNT), while rats in group VI were not traumatized and were not treated (NTNT). The administration of treatment commenced 30 minutes after the occurrence of traumatic brain injury (TBI) and persisted for 21 consecutive days. The fruit was extracted using water, while HPLC quantified the phytochemicals. Neurological severity scores and novel object discrimination tests were carried out. The histological appearance of the brain tissue was also evaluated. The tamarind fruit pulp extract showed the characteristic peaks of rutin flavonoids, caffeic acids and tannic acids in the HPLC analysis. Tamarind fruit pulp extract improved neurological scores and memory function in the treated groups compared to group V, which did not show improvement. Histological results showed few lesions in the treated groups, while in the TNT group, massive and diffused lesions were observed. This study has shown that Tamarindus indica fruit contains neuro-therapeutic substances which may benefit TBI patients by improving their neurological and memory function.

Damned if she does: The subordinate male target hypothesis and discrimination of social dominant female minority members.

According to the subordinate male target hypothesis (SMTH), racism is based on an ethnicity-by-gender interaction, with a stronger link between experiencing racist discrimination and subordinate or dominant ethnic group status for men compared to women. This study reevaluates the SMTH, originally focused on objective discrimination, by applying it to self-reported active harm as a theoretically derived measure of racist discrimination and by exploring interindividual differences in female ethnic minority members' discriminatory experiences. We proposed that social dominance orientation (SDO) among female ethnic minorities would influence SMTH predictions. We tested this using multiple linear regression analyses among a sample of New Zealand Europeans as the majority in New Zealand and non-Europeans as the minority. As hypothesized, male non-Europeans reported disproportionally more active harm than female non-Europeans. Unexpectedly, not only female but also male, non-Europeans high in SDO reported more active harm than non-Europeans low in SDO. Therefore, applied to self-reported racist experiences, oppression of ethnic minorities is driven by interindividual differences rather than by gender. Together, these findings provide evidence that the SMTH cannot be unreservedly extended to reports of racist discrimination and that other processes may underlie these subjective experiences of discrimination that need to be considered in more detail.

PERCEIVED AGE DISCRIMINATION IN THE SECOND HALF OF LIFE: AN EXAMINATION OF AGE, PERIOD, AND COHORT EFFECTS

Abstract The present study aimed to examine variations in perceived age discrimination in the second half of life. Perceived age discrimination is different from objective age discrimination as it depends on respondents’ recognition, acknowledgement, and willingness to report the discriminatory event. We adopt a comprehensive approach which examines whether perceived age discrimination varies by age (chronological time from birth), period (when data were collected), or cohort (a group of people with shared life events experienced at a similar age) across gender and ethnic origin. Relying on psychosocial data from Health and Retirement Survey between 2006 and 2018. Our findings show that perceived age discrimination increases with age. In addition, older cohorts of women are more likely to report perceived age discrimination, whereas White women born around 1940 are less likely to report perceived age discrimination. White men, in contrast show a slight increase in perceived discrimination. Results are discussed from an intersectional perspective, which considers societal life course events.

Discriminative object tracking by domain contrast

Multi-domain tracking method as a novel learning paradigm improves object tracking by sharing domain information with a common backbone whilst learning private information with domain-specific layers. In that context, each individual video sequence as a specific domain serves for a domain-specific layer. In this paper, we observe an intriguing finding that target features from different domains are highly confused with each other, thus having weak discriminative ability, for lack of domain interaction. To this end, we propose a simple yet effective domain interaction training paradigm called domain contrast to boost discriminative object features by effectively using amounts of instances from all the domains in two novel aspects: (1) a light-weight memory-saving training algorithm is proposed to solve the “out-of-the-memory” problem, which paves the way to couple with previous multi-domain trackers, and (2) a composite class-balanced loss is explored to tackle a more practical imbalanced problem, which not only involves the usual class imbalance problem but also accounts for the case of the totally mere negative instances. Experiments on multiple popular tracking benchmark datasets show that our mechanism consistently achieves the tracking performance gain of both base multi-domain tracker and its real-time variant thereof, without any other changes made on the original network.

Brain dynamics during mnemonic discrimination in preclinical and prodromal Alzheimer’s disease

AbstractBackgroundMnemonic discrimination of similar lures and targets is emerging as a promising marker of subtle memory decline in early Alzheimer’s disease (AD) due to its reliance on regions most affected by AD pathology. Tau pathology initially targets medial temporal lobe (MTL) regions in an anterior‐temporal (AT) brain network, which is preferentially involved in object processing, before it can be found in the posterior‐medial (PM) network associated with scene processing. Tests of object and scene mnemonic discrimination may therefore serve as behavioural indicators of the functional integrity of AT and PM regions, respectively. We investigated behavioural and neural changes in mnemonic discrimination across the AD spectrum.Method201 DELCODE cohort participants underwent fMRI scanning during an object‐scene mnemonic discrimination task at four time points from baseline to three‐year follow‐up. At baseline, participants were split by diagnostic group (cognitively unimpaired = CU, subjective cognitive decline = SCD, mild cognitive impairment = MCI, dementia of the Alzheimer’s type = DAT; Table 1). We identified brain regions showing lower activity to repeated targets but not similar lures, indicative of successful pattern separation of overlapping memory representations.ResultThese findings present cross‐sectional baseline data. Behavioural data showed a main effect of group but no group by category (objects, scenes) interaction. CUs and SCDs outperformed patients across domains and CUs scored higher than SCDs only in scene discrimination. Amyloid positive and negative CUs and SCDs did not differ in performance. Successful mnemonic discrimination engaged hippocampus, parahippocampal cortex, precuneus, angular gyrus, and inferior frontal regions for both objects and scenes. Correct object discrimination was additionally associated with amygdala and transentorhinal cortical activity. Reduced brain activation in regions of the anterior‐temporal and posterior‐medial network during successful object and scene mnemonic discrimination, respectively, was found in patient groups (MCI+DAT) compared to cognitively normal (CU+SCD) older adults.ConclusionOur findings lend further support to the utility of mnemonic discrimination as indicator of memory impairment in early AD. Object and scene tasks engage partially dissociable brain networks and individuals with cognitive impairment show reduced neural pattern separation effects. These findings lay the groundwork for future longitudinal analyses to track these changes based on clinical and biomarker status.

Greater age‐related cognitive decline in perceptual discrimination of objects than scenes: A cross‐cultural study

AbstractBackgroundAccording to the “PMAT” framework, two distinct cortico‐hippocampal networks – the posterior‐medial (PM) and anterior‐temporal (AT) networks – are involved in the processing of scene and object content respectively (Ranganath & Ritchey, 2012). It has been suggested that aging predominantly affects the AT network (Berron et al., 2018), and recent studies show evidence for early impairment in object relative to scene mnemonic discrimination with age (Reagh et al., 2016; Gusten et al., 2021). Age‐dependent impairments in object discrimination even exist outside the realm of episodic memory tasks, such as in perceptual oddity tasks (Graham et al., 2010). We aim to extend these findings with i) a digital Oddity perceptual discrimination task, and ii) a cross‐cultural sample.MethodWe tested healthy participants in the UK (n = 141) and India (n = 149) aged between 18‐70 years on a novel multi‐category Oddity discrimination task including object and scene categories. Using a combined speed‐accuracy measure, Inverse Efficiency Score (IES), we applied a mixed‐effects modelling approach to analyse the age trajectory of performance across categories.ResultsWe found a significant decline in performance with age across stimulus categories and cultures, and object performance showed a significantly steeper decline than scenes in both cultures.ConclusionThese findings are amongst the first to indicate that the Oddity discrimination task is sensitive and specific to age‐related cognitive decline across cultures. Notably, these findings support the idea that the object/ AT network is more sensitive to age (potentially reflecting tau pathology) than the scene/ PM network (Berron et al., 2020). Therefore, perceptual discrimination tasks may be a promising cross‐cultural assessment tool for the early detection and differential diagnosis of patterns of age‐related cognitive decline.

The AppNLGF mouse model of amyloid pathology: age‐related behavioural characterisation

AbstractBackground2nd generation knock‐in models of amyloid pathology have been generated to overcome the non‐specific effects of the overexpression of human APP mutation in mice (Saito et al., 2014). AppNLGF mice are a relatively new model that expresses human mutant APP at endogenous levels without cellular artefacts noted in transgenic mice (Sasaguri et al., 2022). The key question of how this more restricted expression of human APP mutations influences behaviour in mice has yet to be fully understood. The aim of this study was to establish the profile of memory and anxiety processes in AppNLGF mice with age.MethodsTo monitor the effect of progressive Aβ deposition, AppNLGF mice underwent testing at 4‐6, 8‐10, 12‐14, and 22‐23 months old. Behavioural tests were selected to interrogate key neuronal networks affected in AD. Spatial recognition memory was assessed using the Object Location Test (OLT). Spatial working memory was tested using Spontaneous Alternation (SA) testing. Anxiety‐like behaviours were observed with Open Field (OF) and Elevated Plus Maze (EPM) tests.ResultsDuring OLT, AppNLGF mice showed no significant difference in object discrimination compared to control animals at all ages tested. However, at 8‐10 and 22‐23m/o, AppNLGFs were unable to discriminate the object in the novel location. Furthermore, AppNLGFs showed intact alternation in SA testing until 22‐23m/o. Interestingly, AppNLGFs exhibited decreased inner zone exploration in OF testing at 22‐23m/o, but consistently showed substantial preference for open arms in the EPM test from as early as 4m/o.ConclusionThe AppNLGF model shows subtle cognitive deficits that are most apparent at an advanced age. The severity of the deficits is relatively mild compared to transgenic models assessed on similar tasks. This is probably related to the absence of overexpression‐related artefacts and may speak to the more subtle effects of amyloid on synaptic plasticity. These mice also lack key cellular AD hallmarks such as tau hyperphosphorylation and cell loss. Nevertheless, AppNLGFs robustly display substantial early anxiolytic‐like behaviour in the EPM test, and anxiogenic‐like behaviour in the OF test at an advanced age, suggesting that amyloid pathology may contribute to behavioural and neuropsychiatric changes.

Comfort Therapy with Companionship Improves Learning Memory and Reduces Hyperalgesia in Sickle Mice

Chronic and unpredictable acute pain coexist with cognitive dysfunction in sickle cell disease (SCD). Managing chronic pain is crucial, though efforts are complicated by both physiological and affective contributors to pain. The affective state, which is severely impacted by the environment including stigma and social isolation, plays a substantial role in pain perception and may further exacerbate psychological factors. Therefore, we examined if companionship and social isolation influence cognitive function and pain in SCD. In the present study, we examined the effects of social isolation on hyperalgesia, learning memory, and anxiety in humanized transgenic homozygous BERK mouse model of SCD. At ~3.5 months of age, BERK male mice were paired with age-matched female BERK companions (1 male + 1 female/cage; called companion group) and another group of BERK males were placed into isolation (one male/cage; called isolation group) for ~14 months. Isolation and companion groups were maintained in general housing conditions in the same room. Both groups were assessed for body weight, hyperalgesia and cognitive behaviors. Mechanical hyperalgesia was assessed by counting paw withdrawal frequency (PWF) following 10 applications of a 1.0 g von Frey monofilament (Semmes Weinstein, Stoelting, Wood Dale, IL, USA) to the plantar surface of the hind paw. Cold hyperalgesia was assessed following placement of mice onto a 4°C cold plate chamber (Stoelting) and by measuring PWF over a 2-minute period. Musculoskeletal hyperalgesia was measured by testing grip force of each mouse's peak forepaw grip to a computerized grip force meter (SA Maier, Milwaukee, WI, USA) and normalizing by body weight. The cognitive behavioral assessment consisted of an open field test to assess anxiety-related behaviors, which measured time spent and number of crosses through a central region. Working memory was assessed using novel object recognition and novel placement recognition tests to measure discrimination of novel objects and location. Data are shown as mean ± SEM and comparisons between companionship and isolation groups were performed using unpaired, two-tailed T-tests to determine statistical significance (P<0.05) We observed a significant increase in the body weights of males housed with females ( Companion group) compared to males in isolation (p<0.05). It is likely that males living with a companion were eating more compared to those under isolation. Isolated male sickle mice experience greater hyperalgesia than male mice housed with a female companion. Isolation induces greater PWF to mechanical (P<0.05) and cold (P<0.005) stimuli, indicating exacerbated hypersensitivity to mechanical and cold stimuli. Spontaneous, non-evoked musculoskeletal hyperalgesia was not significantly altered by isolation (P>0.05). Isolation also appeared to impact working and learning memory. Novel object recognition and novel object placement tests revealed reduced discrimination ratios in isolated mice (~25%, P<0.05 and ~25%, P<0.05), which are suggestive of impaired learning and memory function compared to mice with companionship. Open field test showed no significant difference (P>0.05) between isolation and companionship groups in time spent and number of crosses in the open field center area, indicating no detectable differences in anxiety-like behaviors. According to cognitive tests, mice in companionship showed greater novel object and place detection and greater cognitive ability compared to those in isolation. For the first time we show the causal role of social isolation on hyperalgesia and memory in a humanized mouse model of SCD. “ Comfort therapy” with companionship can ameliorate the perception of pain through the higher brain centres. It is also likely that mice in companionship eat more as evinced by a significant increase in their weight, which may provide additional comfort and meet the increased nutritional/calorie demands of SCD due to altered metabolism. Our observations suggest that “ comfort therapy” with improved living conditions and reducing isolation and loneliness may improve pain-related outcomes in SCD.

Dark-Channel Soft-Constrained and Object-Perception-Enhanced Deep Dehazing Networks Used for Road Inspection Images.

Haze seriously affects the visual quality of road inspection images and contaminates the discrimination of key road objects, which thus hinders the execution of road inspection work. The basic assumptions of the classical dark-channel prior are not suitable for road images containing light-colored lane lines and vehicles, while typical deep dehazing networks lack physical model interpretability, and they focus on global dehazing effects, neglecting the preservation of object features. For this reason, this paper proposes a Dark-Channel Soft-Constrained and Object-Perception-Enhanced Deep Dehazing Network (DCSC-OPE-Net) for the information recovery of road inspection images. The network is divided into two modules: a dark-channel soft-constrained dehazing module and a near-view object-perception-enhanced module. Unlike the traditional dark-channel algorithms that impose strong constraints on dark pixels, a dark-channel soft-constrained loss function is constructed to ensure that the features of light-colored vehicles and lane lines are effectively maintained. To avoid resolution loss due to patch-based dark-channel processing for image dehazing, a resolution enhancement module is used to strengthen the contrast of the dehazed image. To autonomously perceive and enhance key road features to support road inspection, edge enhancement loss combined with a transmission map is embedded into the network to autonomously discover near-view objects and enhance their key features. The experiments utilize public datasets and real road inspection datasets to validate the performance of the proposed DCSC-OPE-Net compared with typical networks using dehazing evaluation metrics and road object recognition metrics. The experimental results demonstrate that the proposed DCSC-OPE-Net can obtain the best dehazing performance, with an NIQE score of 4.5 and a BRISQUE score of 18.67, and obtain the best road object recognition results (i.e., 83.67%) among the comparison methods.

Young children’s subjective and objective thresholds and emergent processes of visual consciousness using a backward masking task

Visual consciousness studies in humans have primarily focused on adults. However, whether young children’s visual consciousness is similar to or different from that of adults remains unknown. This study examined young children’s and adults’ subjective awareness and objective discrimination for thresholds and emergent processes of visual consciousness in two experiments. In Experiment 1, 20 5–6-year-olds and 20 adults attempted a figure discrimination task using a square or a diamond as the target stimulus and responded, using a two-point scale, to a question on subjective awareness of the target stimulus with stimulus onset asynchronies (SOA) from 20 to 260 ms. In Experiment 2, 31 5–6-year-olds and 16 adults attempted the task and responded, using a four-point scale, to a question on subjective awareness with SOA from 50 to 550 ms. We measured the discrimination accuracy and the awareness scale with SOA and fit them to the sigmoid function. The results showed that the objective accuracy threshold of young children was larger than that of adults. Moreover, young children’s subjective awareness threshold was larger than that of adults in the four-point but not in the two-point scale responses. Finally, there were no age differences in the emergent process of consciousness. This study suggests that the emergent process of consciousness in young children is similar to that in adults, however, the threshold in young children is larger than that in adults.

Formal theories clarify the complex: Generalizing a neural process account of the interaction of visual exploration and word learning in infancy.

The interaction of visual exploration and auditory processing is central to early cognitive development, supporting object discrimination, categorization, and word learning. Research has shown visual-auditory interactions to be complex, created from multiple processes and changing over multiple timescales. To better understand these interactions, we generalize a formal neural process model of early word learning to two studies examining how words impact 9- to 22-month-olds' attention to novelty. These simulations clarify the origin and nature of previously demonstrated effects of labels on visual exploration and the basis of mutual exclusivity effects in word learning. We use our findings to discuss key questions for this special section: what makes a good theory and how should formal theories interface with empirical paradigms and findings?

Comparison of reflective band (Vis, NIR, SWIR, eSWIR) performance in daytime reduced illumination conditions

Daytime low-light conditions such as overcast, dawn, and dusk pose a challenge for object discrimination in the reflective bands, where the majority of illumination comes from reflected solar light. In reduced-illumination conditions, the sensor signal-to-noise ratio can suffer, inhibiting range performance for detecting, recognizing, and identifying objects of interest. This performance reduction is more apparent in the longer wavelengths where there is less solar light. Range performance models show a strong dependence on cloud type and thickness, as well as time of day across the reflective wavebands. Through an experimental and theoretical analysis of a passive sensitivity- and resolution-matched testbed, we compare Vis (0.4–0.7 µm), NIR (0.7–1 µm), SWIR (1–1.7 µm), and eSWIR (2–2.5 µm) to assess the limiting cases in which reduced illumination inhibits range performance. The time during dawn and dusk is brief yet does show significant range performance reduction for SWIR and eSWIR. Under heavy cloud cover, eSWIR suffers the most at range due to a low signal-to-noise ratio. In cases of severe reduction in illumination, we propose utilizing active illumination or the emissive component of eSWIR to improve the signal-to-noise ratio for various discrimination tasks.

A Waste Classification model in Low-illumination scenes based on ConvNeXt

Waste classification is an essential part of environmental pollution management in modern society. Object detection is an accurate and efficient way to classify waste, which is conducive to recycling resources. However, due to low object discriminability, existing waste classification models cannot classify waste in low-illumination scenes. A waste classification model, Dark-Waste, is designed to classify wastes in a low-illumination scenario. Firstly, to solve the scarcity of training data, an efficient and low-cost Illumination Conversion method is proposed to generate the low-light image. Secondly, the improved ConvNeXt network is combined with YOLOv5 to accurately and efficiently classify waste. Finally, we validated the model on a self-built dataset in real scenarios. The experimental results show that Dark-Waste achieves the best detection performance in low-illumination scenes. The Dark-Waste provides a new approach to waste management in complex environments and effectively contributes to the sustainable development of the urban ecological environment.

Pro-social and pro-cognitive effects of LIT-001, a novel oxytocin receptor agonist in a neurodevelopmental model of schizophrenia

Social and cognitive dysfunctions are the most persistent symptoms of schizophrenia. Since oxytocin (OXT) is known to play a role in social functions and modulates cognitive processes, we investigated the effects of a novel, nonpeptide, selective OXT receptor agonist, LIT-001, in a neurodevelopmental model of schizophrenia. Administration of methylazoxymethanol acetate (MAM; 22 mg/kg) on the 17th day of rat pregnancy is known to cause developmental disturbances of the brain, which lead to schizophrenia-like symptomatology in the offspring. Here, we examined the effects of acutely administered LIT-001 (1, 3, and 10 mg/kg) in MAM-exposed males and females on social behaviour, communication and cognition. We report that MAM-treated adult male and female rats displayed reduced social behaviour, ultrasonic communication and novel object recognition test performance. LIT-001 partially reversed these deficits, increasing the total social interaction time and the number of ‘positive’, highly-modulated 50 kHz ultrasonic calls in male rats. The compound ameliorated MAM-induced deficits in object discrimination in both sexes. Present results confirm the pro-social activity of LIT-001 and demonstrate its pro-cognitive effects following acute administration.

HAMIL: High-Resolution Activation Maps and Interleaved Learning for Weakly Supervised Segmentation of Histopathological Images.

Semantic segmentation of histopathological images is important for automatic cancer diagnosis, and it is challenged by time-consuming and labor-intensive annotation process that obtains pixel-level labels for training. To reduce annotation costs, Weakly Supervised Semantic Segmentation (WSSS) aims to segment objects by only using image or patch-level classification labels. Current WSSS methods are mostly based on Class Activation Map (CAM) that usually locates the most discriminative object part with limited segmentation accuracy. In this work, we propose a novel two-stage weakly supervised segmentation framework based on High-resolution Activation Maps and Interleaved Learning (HAMIL). First, we propose a simple yet effective Classification Network with High-resolution Activation Maps (HAM-Net) that exploits a lightweight classification head combined with Multiple Layer Fusion (MLF) of activation maps and Monte Carlo Augmentation (MCA) to obtain precise foreground regions. Second, we use dense pseudo labels generated by HAM-Net to train a better segmentation model, where three networks with the same structure are trained with interleaved learning: The agreement between two networks is used to highlight reliable pseudo labels for training the third network, and at the same time, the two networks serve as teachers for guiding the third network via knowledge distillation. Extensive experiments on two public histopathological image datasets of lung cancer demonstrated that our proposed HAMIL outperformed state-of-the-art weakly supervised and noisy label learning methods, respectively. The code is available at https://github.com/HiLab-git/HAMIL.

Transmembrane protein 184B (TMEM184B) promotes expression of synaptic gene networks in the mouse hippocampus

In Alzheimer’s Disease (AD) and other dementias, hippocampal synaptic dysfunction and loss contribute to the progression of memory impairment. Recent analysis of human AD transcriptomes has provided a list of gene candidates that may serve as drivers of disease. One such candidate is the membrane protein TMEM184B. To evaluate whether TMEM184B contributes to neurological impairment, we asked whether loss of TMEM184B in mice causes gene expression or behavior alterations, focusing on the hippocampus. Because one major risk factor for AD is age, we compared young adult (5-month-old) and aged (15-month-old) wild type and Tmem184b-mutant mice to assess the dual contributions of age and genotype. TMEM184B loss altered expression of pre- and post-synaptic transcripts by 5 months and continued through 15 months, specifically affecting genes involved in synapse assembly and neural development. Wnt-activated enhancer elements were enriched among differentially expressed genes, suggesting an intersection with this pathway. Few differences existed between young adult and aged mutants, suggesting that transcriptional effects of TMEM184B loss are relatively constant. To understand how TMEM184B disruption may impact behaviors, we evaluated memory using the novel object recognition test and anxiety using the elevated plus maze. Young adult Tmem184b-mutant mice show normal object discrimination, suggesting a lack of memory impairment at this age. However, mutant mice showed decreased anxiety, a phenotype seen in some neurodevelopmental disorders. Taken together, our data suggest that TMEM184B is required for proper synaptic gene expression and anxiety-related behavior and is more likely to be linked to neurodevelopmental disorders than to dementia.