Cross-Domain Human Activity Recognition Using Low-Resolution Infrared Sensors.

This paper investigates the feasibility of cross-domain recognition for human activities captured using low-resolution 8 × 8 infrared sensors in indoor environments. To achieve this, a novel prototype recurrent convolutional network (PRCN) was evaluated using a few-shot learning strategy, classifying up to eleven activity classes in scenarios where one or two individuals engaged in daily tasks. The model was tested on two independent datasets, with real-world measurements. Initially, three different networks were compared as feature extractors within the prototype network. Following this, a cross-domain evaluation was conducted between the real datasets. The results demonstrated the model's effectiveness, showing that it performed well regardless of the diversity of samples in the training dataset.

CAPTURE-24: A large dataset of wrist-worn activity tracker data collected in the wild for human activity recognition.

Existing activity tracker datasets for human activity recognition are typically obtained by having participants perform predefined activities in an enclosed environment under supervision. This results in small datasets with a limited number of activities and heterogeneity, lacking the mixed and nuanced movements normally found in free-living scenarios. As such, models trained on laboratory-style datasets may not generalise out of sample. To address this problem, we introduce a new dataset involving wrist-worn accelerometers, wearable cameras, and sleep diaries, enabling data collection for over 24 hours in a free-living setting. The result is CAPTURE-24, a large activity tracker dataset collected in the wild from 151 participants, amounting to 3883 hours of accelerometer data, of which 2562 hours are annotated. CAPTURE-24 is two to three orders of magnitude larger than existing publicly available datasets, which is critical to developing accurate human activity recognition models.

BodyFlow: An Open-Source Library for Multimodal Human Activity Recognition.

Human activity recognition is a critical task for various applications across healthcare, sports, security, gaming, and other fields. This paper presents BodyFlow, a comprehensive library that seamlessly integrates human pose estimation and multiple-person estimation and tracking, along with activity recognition modules. BodyFlow enables users to effortlessly identify common activities and 2D/3D body joints from input sources such as videos, image sets, or webcams. Additionally, the library can simultaneously process inertial sensor data, offering users the flexibility to choose their preferred input, thus facilitating multimodal human activity recognition. BodyFlow incorporates state-of-the-art algorithms for 2D and 3D pose estimation and three distinct models for human activity recognition.

Artificial light at night reveals hotspots and rapid development of industrial activity in the Arctic.

Climate warming enables easier access and operation in the Arctic, fostering industrial and urban development. However, there is no comprehensive pan-Arctic overview of industrial and urban development, which is crucial for the planning of sustainable development of the region. In this study, we utilize satellite-derived artificial light at night (ALAN) data to quantify the hotspots and the development of light-emitting human activity across the Arctic from 1992 to 2013. We find that out of 16.4 million km2 analyzed a total area of 839,710 km2 (5.14%) is lit by human activity with an annual increase of 4.8%. The European Arctic and the oil and gas extraction regions in Russia and Alaska are hotspots of ALAN with up to a third of the land area lit, while the Canadian Arctic remains dark to a large extent. On average, only 15% of lit area in the Arctic contains human settlement, indicating that artificial light is largely attributable to industrial human activity. With this study, we provide a standardized approach to spatially assess human industrial activity across the Arctic, independent from economic data. Our results provide a crucial baseline for sustainable development and conservation planning across the highly vulnerable Arctic region.

Habitat Diversity Mitigates the Impacts of Human Pressure on Stream Biodiversity.

Recent decades have witnessed substantial changes in freshwater biodiversity worldwide. Although research has shown that freshwater biodiversity can be shaped by changes in habitat diversity and human-induced pressure, the potentials for interaction between these drivers and freshwater biodiversity at large spatial extents remain unclear. To address these issues, we employed a spatially extensive multitrophic fish and insect database from 3323 stream sites across the United States, to investigate the ability of habitat diversity to modulate the effect of human pressure on the richness and abundance of fish and insects. We found evidence that high levels of habitat diversity were associated with increased richness and abundance of fish and insects (including whole-assemblage and individual trophic guilds). We also show that the effects of human pressure on the richness and abundance of fish and insects tend to become positive at high levels of habitat diversity. Where habitat diversity is low, human pressure strongly reduces insect richness and abundance, whereas these reductions are attenuated at high levels of habitat diversity. Structural equation modeling revealed that human pressure reduced habitat diversity, indirectly negatively affecting the richness and abundance of fish and insects. These findings illustrate that, in addition to promoting greater fish and insect biodiversity, habitat diversity may mitigate the deleterious effects of human pressures on these two stream assemblages. Overall, our study suggests that maintaining high levels of habitat diversity is a useful way to protect freshwater biodiversity from ongoing increases in human pressure. However, if human pressures continue to increase, this will reduce habitat diversity, further threatening stream assemblages.

[Impact of Climate Change and Human Activities on the Ecological Sensitivity of the Yangtze River Economic Belt].

The increasing climate change and human activities exert their influence on the ecological sensitivity of the region individually and interactively. Therefore, a clear understanding of the impact of climate change and human activities on ecological sensitivity will enhance the resilience of the regional ecological environment and the level of sustainable development. This study took the Yangtze River Economic Belt, the first demonstration zone of China's ecological civilization construction, as the research object. Based on the meteorological, remote sensing, and statistical data of 130 cities in the whole region from 2001 to 2021, an index system of climate change, human activities, and ecological sensitivity was constructed. Response surface methodology （RSM） was used to explore the effects of climate and anthropogenic single factors and interactions on the ecological sensitivity in each region. The results showed that： ① The ecological sensitivity value of the belt fluctuated and rose in time, rising by 2.2% from 2001 to 2021. In terms of space, the overall spatial distribution was "high in the north and low in the south." In 2021, the proportion of severely and extremely sensitive cities in the Yangtze River Economic Belt reached nearly 50%. ② For a single factor, the distribution of the effect of the same factor had certain characteristics： The areas where the single factors of economic development, rainfall, and temperature had a positive impact on the ecological sensitivity were concentrated in the areas with higher or faster economic development, along and south of the Yangtze River. For the interaction factors, the effect of 78.6% of the factors on the ecological sensitivity was negative interaction, and the change of one factor level would change the direction of the effect of the other factor on the regional sensitivity. ③ The comprehensive ecological management area of the Yangtze River Economic Belt was divided based on the ecological sensitivity and climate sensitivity. The governance areas that needed priority improvement were clustered within the three urban agglomerations and their northern adjacent areas, which meant that the ecological sensitivity and climate sensitivity of a city had spillover effects. This study is expected to provide inspiration for the economic zone and even the national and global efforts in the field of regional ecological governance.

Impact of Human Activities on Woody Vegetation in Gallery Forests in the Mandara Mountains (Far North, Cameroon).

This study was conducted in the Mandara Mountains in Cameroon and aimed to assess the effects of human activities on woody vegetation in gallery forests, based on floristic inventories and observations made by the government. Firstly, the inventories were carried out in 150 plots of 1000 m2 each, installed on the banks of watercourses following the band of plant formations. In each plot, woody species were counted and those showing at least one sign of degradation were noted. Secondly, the survey was conducted in 18 administrative structures made up of delegations (MINFOF, MINADER, MINEPDED, and MINEPIA) and town halls. One hundred woody species, grouped into 63 genera and 30 families, have been inventoried, in which 45 species showed at least one sign of damage caused by human being. The species most affected are Anogeissus leiocarpus (67 stems), Azadirachta indica (46 stems), Diospyros mespiliformis (43 stems), Acacia albida (42 stems), Andira inermis (30 stems), Acacia sieberiana (23 stems), Khaya senegalensis (19 stems), Ficus sycomorus (13 stems), and Acacia polyacantha (10 stems). The most recurrent activity in the gallery forests is pruning (212 stems), followed by cutting (93 stumps), then picking (71 individuals). However, there are fewer debarked trees (11) and trees with fire trail (6). According to the responses provided, logging (77.78%), agriculture (72.22%), population growth (44.44%), grazing (33.33%), and bush fires (33.33%) are the main causes of the degradation of plant formations in the Mandara Mountains. These main factors could have a negative impact on biodiversity if appropriate integrated management measures are not taken. To maintain these vital ecosystems, an integrated management plan must be put in place, limiting human activities to a minimum.

Early emergence and determinants of human-induced Walker circulation weakening.

The Walker circulation is projected to slow down in response to greenhouse gas warming. However, detecting the impact of human activities on changes in the Walker circulation is challenging due to the significant influence of internal variability. Here, based on ensembles of multiple climate models from the Coupled Model Intercomparison Project Phase 6 (CMIP6), we show evidence that the emergence of the human-induced weakening of Walker circulation tends to occur earlier in the middle-upper troposphere than at the surface. This earlier emergence is attributed to a more pronounced initial weakening response of the middle-upper tropospheric Walker circulation to atmospheric CO2 radiative forcing. We further reveal that the emergence time of a weaker Walker circulation varies across models. This intermodel spread is governed by an ocean thermostat that operates by modulating the zonal sea surface temperature gradient over the tropical Indo-Pacific region. Our findings address the key question of whether and how to detect human-induced large-scale atmospheric circulation changes and provide valuable insights for assessing the associated risks.

The quantifying, mapping, and risk analysis of human-related stressors in the high seas.

Objectives: Marine biodiversity and ecosystem services in the high seas are threatened by numerous stress factors caused by human activities, including global shipping, high-sea fishing, marine plastic pollution, and anthropogenic climate change. Socioeconomic factors are one of the criteria for the establishment of area-based management tools in the high seas for marine biodiversity conservation beyond national jurisdiction. The aim of the work is to propose a spatiotemporal approach to identify risks from marine human activities and recommendations for high seas governance. Methods: Data related to human activities from 2014 to 2022 were used to calculate the distribution and changes of human-related stressors, and the risk to marine biodiversity in the high seas caused by human activities. Results: The North Atlantic, Philippine Sea, Arabian Sea, Bay of Bengal, and East Central Atlantic show high and increasing intensities of human-related stressors, and are therefore particularly at need for the protection and conservation of marine biodiversity. Risks from human activities vary within the marine areas that are prioritized for biodiversity protection. The study recommends that the designation of high seas protected areas should take into account the types of risks to which the different marine areas are exposed, and that the high seas protected areas should be established gradually. At the same time, appropriate management measures should be formulated according to the intensity of human activities in the different marine areas. Conclusions: Quantifying and classifying the risk from human-related stressors could help identify solution for the protection and conservation and facilitate the marine spatial planning, establishment area based management tools, including marine protected areas in the high seas.

Human activities shape important geographic differences in fish mercury concentration levels.

Fish consumption is a major route of human exposure to mercury (Hg), yet limited understanding of how anthropogenic activities drive geographic variations in fish Hg worldwide hinders effective Hg pollution management. Here we characterized global geographic variations in total Hg (THg) and methylmercury (MeHg), compared THg and MeHg levels between the United States and China, and used a structural equation model to link the geographic variability of MeHg in fish to human activities. Despite previously reported higher Hg emissions in China, Chinese fish have lower THg and MeHg levels than fish in the United States owing to a lower trophic magnification slope, shortened food chains and shorter fish lifespans. The structural equation model revealed strong impacts of human activities on MeHg levels in fish. In the future, China may face elevated MeHg levels in fish with the ongoing recovery of food web ecology, highlighting the importance of local policies.

A Novel Active Learning Framework for Cross-Subject Human Activity Recognition from Surface Electromyography.

Wearable sensor-based human activity recognition (HAR) methods hold considerable promise for upper-level control in exoskeleton systems. However, such methods tend to overlook the critical role of data quality and still encounter challenges in cross-subject adaptation. To address this, we propose an active learning framework that integrates the relation network architecture with data sampling techniques. Initially, target data are used to fine tune two auxiliary classifiers of the pre-trained model, thereby establishing subject-specific classification boundaries. Subsequently, we assess the significance of the target data based on classifier discrepancy and partition the data into sample and template sets. Finally, the sampled data and a category clustering algorithm are employed to tune model parameters and optimize template data distribution, respectively. This approach facilitates the adaptation of the model to the target subject, enhancing both accuracy and generalizability. To evaluate the effectiveness of the proposed adaptation framework, we conducted evaluation experiments on a public dataset and a self-constructed electromyography (EMG) dataset. Experimental results demonstrate that our method outperforms the compared methods across all three statistical metrics. Furthermore, ablation experiments highlight the necessity of data screening. Our work underscores the practical feasibility of implementing user-independent HAR methods in exoskeleton control systems.

Assessing the impact of human activities on ecosystem asset dynamics in the Yellow River Basin from 2001 to 2020.

The intensification of human activities in the Yellow River Basin has significantly altered its ecosystems, challenging the sustainability of the region's ecosystem assets. This study constructs an ecosystem asset index for the period from 2001 to 2020, integrating it with human footprint maps to analyze the temporal and spatial dynamics of ecosystem assets and human activities within the basin, as well as their interrelationships. Our findings reveal significant improvement of ecosystem assets, mainly attributed to the conversion of farmland back into natural habitats, resulting in a 15,994 km2 increase in ecological land use. Notably, 45.88% of the basin has experienced concurrent growth in both human activities and ecosystem assets, with ecosystem assets expanding at a faster rate (22.61%) than human activities (17.25%). Areas with high-quality ecosystem assets are expanding, in contrast to areas with intense human activities, which are facing increased fragmentation. Despite a global escalation in threats from human activities to ecosystem assets, the local threat level within the Yellow River Basin has slightly diminished, indicating a trend towards stabilization. Results highlight the critical importance of integrating spatial and quality considerations into restoration efforts to enhance the overall condition of ecosystem assets, especially under increasing human pressures. Our work assesses the impact of human activities on the dynamics of ecosystem assets in the Yellow River Basin from 2001 to 2020, offering valuable insights for quality development in the region, may provide a scientific basis for general watershed ecological protection and sustainable management in a region heavily influenced by human activity but on a path to recovery.

Human activities exacerbate river network degrading in the Qinhuai River basin.

Rivers are undergoing significant changes under the pressures of natural processes and human activities. However, characterizing and understanding these changes over the long term and from a spatial perspective have proven challenging. This paper presents a novel framework featuring twelve indicators that combine geometric and spatial structures for evaluating changes in river network patterns. Through global principal component analysis, these indicators were integrated into a comprehensive river network pattern index (RNP). Employing Pearson correlation analysis, geographically weighted regression, geographic detector models, and the Shapley Value, the study quantitatively analyzed various stressors' impacts and relative contributions on river network changes from the 1960s to 2015s. The results showed a clear trend of degradation over time, particularly with frequency and density declining by 57 % and 48 %, respectively. The changes across subbasins varied temporally and spatially, with the 1980s emerging as a significant temporal hotspot and six spatial hotspots identified among twenty subbasins. The analysis showed that agriculture was significantly negatively associated with RNP, while the relationship between urbanization and RNP was inverted N-shaped. To address the negative effects of human activities, a shift from uniform management approaches is crucial. In agricultural areas, adopting more intensive farming practices could help mitigate negative impacts on RNP. For highly urbanized regions, city planning should consider the interactions between urbanization and other factors affecting RNP. Overall, incorporating an understanding of RNP's spatial-temporal dynamics and driving factors into spatial planning is critical for creating effective and sustainable management strategies for human-river interactions.

Cumulative human impacts on global marine fauna highlight risk to biological and functional diversity.

Anthropogenic stressors to marine ecosystems from climate change and human activities increase extinction risk of species, disrupt ecosystem integrity, and threaten important ecosystem services. Addressing these stressors requires understanding where and to what extent they are impacting marine biological and functional diversity. We model cumulative risk of human impact upon 21,159 marine animal species by combining information on species-level vulnerability and spatial exposure to a range of anthropogenic stressors. We apply this species-level assessment of human impacts to examine patterns of species-stressor interactions within taxonomic groups. We then spatially map impacts across the global ocean, identifying locations where climate-driven impacts overlap with fishing, shipping, and land-based stressors to help inform conservation needs and opportunities. Comparing species-level modeled impacts to those based on marine habitats that represent important marine ecosystems, we find that even relatively untouched habitats may still be home to species at elevated risk, and that many species-rich coastal regions may be at greater risk than indicated from habitat-based methods alone. Finally, we incorporate a trait-based metric of functional diversity to identify where impacts to functionally unique species might pose greater risk to community structure and ecosystem integrity. These complementary lenses of species, function, and habitat provide a richer understanding of threats to marine biodiversity to help inform efforts to meet conservation targets and ensure sustainability of nature's contributions to people.

Improving Human Activity Recognition With Wearable Sensors Through BEE: Leveraging Early Exit and Gradient Boosting.

Early-exiting has recently provided an ideal solution for accelerating activity inference by attaching internal classifiers to deep neural networks. It allows easy activity samples to be predicted at shallower layers, without executing deeper layers, hence leading to notable adaptiveness in terms of accuracy-speed trade-off under varying resource demands. However, prior most works typically optimize all the classifiers equally on all types of activity data. As a result, deeper classifiers will only see hard samples during test phase, which renders the model suboptimal due to the training-test data distribution mismatch. Such issue has been rarely explored in the context of activity recognition. In this paper, to close the gap, we propose to organize all these classifiers as a dynamic-depth network and jointly optimize them in a similar gradient-boosting manner. Specifically, a gradient-rescaling is employed to bound the gradients of parameters at different depths, that makes such training procedure more stable. Particularly, we perform a prediction reweighting to emphasize current deep classifier while weakening the ensemble of its previous classifiers, so as to relieve the shortage of training data at deeper classifiers. Comprehensive experiments on multiple HAR benchmarks including UCI-HAR, PAMAP2, UniMiB-SHAR, and USC-HAD verify that it is state-of-the-art in accuracy and speed. A real implementation is measured on an ARM-based mobile device.

Fitness consequences of anthropogenic subsidies for a partially migratory wading bird.

Human activities are forcing wildlife to confront selective pressures different from those under which they evolved. In seasonal environments, migration evolved as an adaptation to fluctuating resource availability. Anthropogenic subsidies modify resource dynamics by providing a steady food source that is not subject to seasonality. Globally, many migratory populations are becoming increasingly resident in response to food supplementation. While these population-level shifts are assumed to arise from changing fitness consequences of individual behaviour in response to resource dynamics, these mechanisms are often difficult to quantify and disentangle. Here, we quantified fitness consequences of responses to anthropogenic subsidies in partially migratory wood storks (Mycteria americana) in the heavily urbanized southeastern United States. First, we tested whether individual migratory behaviour is linked to different responses to anthropogenic subsidies. Second, we quantified fitness consequences of these behavioural responses. We found that, in our system, migration and residency are alternative behavioural tactics associated with different responses to food supplementation. In turn, the use of anthropogenic resources alters a fitness component by enhancing nest survival. These results provide a mechanistic examination of how animals may respond to human-modified resource dynamics and how fitness consequences of individual tactics may translate into behavioural shifts at the population level.

A hybrid TCN-GRU model for classifying human activities using smartphone inertial signals.

Recognising human activities using smart devices has led to countless inventions in various domains like healthcare, security, sports, etc. Sensor-based human activity recognition (HAR), especially smartphone-based HAR, has become popular among the research community due to lightweight computation and user privacy protection. Deep learning models are the most preferred solutions in developing smartphone-based HAR as they can automatically capture salient and distinctive features from input signals and classify them into respective activity classes. However, in most cases, the architecture of these models needs to be deep and complex for better classification performance. Furthermore, training these models requires extensive computational resources. Hence, this research proposes a hybrid lightweight model that integrates an enhanced Temporal Convolutional Network (TCN) with Gated Recurrent Unit (GRU) layers for salient spatiotemporal feature extraction without tedious manual feature extraction. Essentially, dilations are incorporated into each convolutional kernel in the TCN-GRU model to extend the kernel's field of view without imposing additional model parameters. Moreover, fewer short filters are applied for each convolutional layer to alleviate excess parameters. Despite reducing computational cost, the proposed model utilises dilations, residual connections, and GRU layers for longer-term time dependency modelling by retaining longer implicit features of the input inertial sequences throughout training to provide sufficient information for future prediction. The performance of the TCN-GRU model is verified on two benchmark smartphone-based HAR databases, i.e., UCI HAR and UniMiB SHAR. The model attains promising accuracy in recognising human activities with 97.25% on UCI HAR and 93.51% on UniMiB SHAR. Since the current study exclusively works on the inertial signals captured by smartphones, future studies will explore the generalisation of the proposed TCN-GRU across diverse datasets, including various sensor types, to ensure its adaptability across different applications.

Vulnerability and driving mechanism of four typical grasslands in China under the coupled impacts of climate change and human activities.

Understanding of how different grasslands types respond to climate change and human activities across different spatial and temporal dimensions is crucial for devising effective strategies to prevent grasslands degradation. In this study, we developed a novel vulnerability assessment model for grasslands that intricately evaluates the combined impact of climate change and human activities. We then applied this model to analyze the vulnerability and driving mechanism of four representative Chinese grasslands to climate change and human activities. Our findings indicate that the vulnerability of the four grasslands would show a pattern of higher in the west and lower in the east under the influence of climate change alone. However, when human activities are factored in, the vulnerability across the four grasslands tends to homogenize, with human activities notably reducing the vulnerability of alpine grasslands in the west and, conversely, increasing the vulnerability of grasslands in the east. Furthermore, our study reveals distinct major environmental drivers of grasslands vulnerability across different regions. The two western alpine grasslands exhibit higher vulnerability to annual mean temperature and isothermality compared to the eastern temperate grasslands, while their vulnerability to precipitation of the coldest quarter is lower than that of the eastern temperate grasslands. These findings are helpful for understanding the multifaceted causes and mechanisms of grasslands degradation, providing a scientific foundation for the sustainable management and conservation of grassland resources.

Exploring the evolution of ecosystem health and sustainable zoning: A perspective based on the contributions of climate change and human activities.

Maintaining ecosystem health (EH) in watersheds is crucial for building a national pattern of ecological security. However, a comprehensive diagnosis of watershed EH and an exploration of its driving mechanisms are still lacking. This study proposed an EH assessment model from a vitality-organization-resilience-service-environment (VORSE) perspective. Taking the Yellow River Basin of Shaanxi Province (YRBS), China, as a research object, the spatiotemporal evolution trend of EH from 2000 to 2020 was quantified. At the same time, we also quantified the respective contributions of climate change (CC) and human activities (HA) to the EH dynamics based on residual analysis. The results showed that EH in the YRBS increased by 11.80 % from 2000 to 2020, and the spatial distribution of the EH was higher in the southern region than in the northern part. At the pixel scale, areas with improving trends accounted for 90.57 % of the YRBS, while 9.43 % deteriorated, with the improving areas mainly in northern Shaanxi and the deteriorating areas in the Guanzhong region. The correlation between the EH and precipitation was primarily positive, while the correlation between the EH and temperature was mainly negative. The residual analysis showed that the contribution rate of CC to EH changes was 78.54 %, while that of HA was 21.46 %, indicating that CC was the dominant driver of EH changes in the YRBS. Specifically, 82.64 % of the improvement in EH was attributed to CC and 17.36 % to HA. Conversely, 65.30 % of the deterioration in EH was attributed to CC and 34.70 % to HA. Furthermore, CC, HA, and CC&HA dominated EH changes in 26.85 %, 3.77 %, and 69.38 % of the YRBS area, respectively. In addition, the Hurst exponent analysis identified six types of future EH development scenarios, each requiring different restoration strategies. This study provides valuable insights for future EH diagnosis, EH restoration efforts, and the formulation of sustainable development goals in other watersheds.