Residual Multi-Attention Classification Network for A Forest Dominated Tropical Landscape Using High-Resolution Remote Sensing Imagery

Summary Baseline data on local status of threatened species are often limited, and alternative information sources such as local ecological knowledge (LEK) have potential to provide conservation insights but require critical evaluation. We assess the usefulness of LEK to generate conservation evidence for the Hainan Peacock-pheasant Polyplectron katsumatae, a poorly known threatened island galliform. Interview surveys in rural communities across eight forested landscapes on Hainan provided a new dataset of sightings of Peacock-pheasants and other galliforms. Fewer respondents had seen Peacock-pheasants compared to other species across most landscapes, although Peacock-pheasant sightings showed significant across-landscape variation, with substantially more total and recent sightings from Yinggeling National Nature Reserve. However, validation of interview data with camera trapping data from Houmiling Provincial Nature Reserve, a landscape with few reported sightings, suggests a more optimistic possible status for Peacock-pheasants, which were detected as frequently as Red Junglefowl Gallus gallus and Silver Pheasant Lophura nycthemera during systematic camera trap placement. Hainan Peacock-pheasant sighting rates might be influenced by various factors (e.g. restricted local access to forests), with absolute abundance possibly greater than expected from limited sightings. Conversely, relative across-landscape abundance patterns from LEK are likely to be valid, as similar detection biases exist across surveyed landscapes.

show abstract

“…2015, Yu et al . 2021). Numerous low-income villages with primarily agricultural-based economies are situated close to each forested area.…”

Section: Methodsmentioning

confidence: 99%

Local ecological knowledge and regional sighting histories of Hainan Peacock-pheasant Polyplectron katsumatae: pessimism or optimism for a threatened island endemic?

Turvey

Ma²,

Zhou³

et al. 2022

Bird Conservation International

View full text Add to dashboard Cite

show abstract

“…Considering that the area of Cassytha filiformis is very small and it is not easy to distinguish it from other vegetation, we used the residual multi-attentive network (ResMANet) proposed in our previous work [38]. ResMANet incorporates a multiscale convolutional neural network (CNN) to extract features at different scales.…”

Section: ) Construction Of the Deep Learning Modelmentioning

confidence: 99%

Multispectral UAV-Based Monitoring of Cassytha Filiformis Invasion in Xisha Islands

Xie,

Wu,

et al. 2024

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

Self Cite

View full text Add to dashboard Cite

Currently, numerous studies have reported that the invasion of Cassytha filiformis has affected both above and below ground communities, resulting in difficulties in the growth of original vegetation. Meanwhile, Cassytha filiformis was observed on the Xisha Islands in recent years which brings up the importance of monitoring its invasion to protect the biodiversity of the island. Nonetheless, to effectively monitor Cassytha filiformis at finer regional scales, there is a pressing need for centimeter-level resolution, a level of precision that current satellite sensors find challenging to attain in a consistent manner. Therefore, we adopted a DJI Phantom 4 Unmanned Aerial Vehicle (UAV) with five multispectral bands and centimeter-level spatial resolution to overcome this problem. An advanced deep learning network is employed to identify the invasion in Xisha Islands for three different time periods. Results show that the area of Cassytha filiformis on Bei Island increased from 211.8 m 2 in April 2020 to 458.6 m 2 in April 2021, and dropped to 112.8 m 2 in July 2021, while that on Ganquan Island changed from 1996.9 m 2 in April 2021 (dry season) to 1275.9 m 2 in July 2021 (wet season). By incorporating climatic indicators, we further found that Cassytha filiformis in both Bei Island and Ganquan Island favors dry climate and its large area invasion in 2021 was possibly caused by a drought event.

show abstract

“…Research in this domain encompasses forest classification, including the classification of landscapes affected by deforestation [58]. Moreso, change detection in vegetation and forest areas enables decision-makers, conservationists, and policymakers to make informed decisions through forest monitoring initiatives [6] and mapping strategies tailored to tropical forests [59].…”

Section: • Forestmentioning

confidence: 99%

A Review and Meta-analysis of Semantic Segmentation Models in Land Use/ Land Cover Mapping

Ajibola,

Cabral

2024

Preprint

View full text Add to dashboard Cite

Recent advancements in deep learning have spurred the development of numerous novel semantic segmentation models for land cover mapping, showcasing exceptional performance in delineating precise boundaries and producing highly accurate land cover maps. However, to date, no systematic literature review has comprehensively examined semantic segmentation models in the context of land cover mapping. This paper addresses this gap by synthesizing recent advancements in semantic segmentation models for land cover mapping from 2017 to 2023, drawing insights on trends, data sources, model structures, and performance metrics based on a review of 106 extracted articles. Our analysis identifies top journals in the field, including MDPI Remote Sensing, IEEE Journal of Selected Topics in Earth Science, and IEEE Transactions on Geoscience and Remote Sensing, IEEE Geoscience and Remote Sensing Letters, ISPRS Journal Of Photogrammetry And Remote Sensing as the leading journals. We find that research predominantly focuses on land cover, urban areas, precision agriculture, environment, coastal areas, and forests. Geographically, 35.29% of the study areas are located in China, followed by USA (11.76%), France (5.88 %), Spain (4%) and others. Sentinel-2, Sentinel-1, and Landsat satellites emerge as the most commonly used data sources. Benchmark datasets such as ISPRS Vaihingen &amp; Potsdam, LandCover.ai, DeepGlobe, and GID datasets are frequently employed. Model architectures predominantly utilize encoder-decoder, and hybrid convolutional neural network-based structures because of their impressive performances, with limited adoption of transformer-based architectures due to its computational complexity issue, and slow convergence speed. Lastly, this paper highlights existing key research gaps in the field to guide future research directions.

show abstract

Residual Multi-Attention Classification Network for A Forest Dominated Tropical Landscape Using High-Resolution Remote Sensing Imagery

Cited by 9 publications

References 33 publications

Local ecological knowledge and regional sighting histories of Hainan Peacock-pheasant Polyplectron katsumatae: pessimism or optimism for a threatened island endemic?

Local ecological knowledge and regional sighting histories of Hainan Peacock-pheasant Polyplectron katsumatae: pessimism or optimism for a threatened island endemic?

Multispectral UAV-Based Monitoring of Cassytha Filiformis Invasion in Xisha Islands

A Review and Meta-analysis of Semantic Segmentation Models in Land Use/ Land Cover Mapping

Contact Info

Product

Resources

About