Skip to main content
Banner [Small]

Test out our new Bento Search

test area
x
# results
shortcut
Sections
HTML elements
Section Tiles
expand
Tile Cover
Mouse
Math Lab
Space
Tile Short Summary
Math Lab Rooms located in the Main Library in rooms 300X and 300Y
expand
Tile Cover
coffee
CC's Coffee House
Space
Tile Short Summary
Located at the first floor of the LSU Main Library.
expand
Tile Cover
People troubleshooting on a computer
Ask Us
Service
Tile Short Summary
Check our FAQs, submit a question using our form, or launch the chat widget to find help.

Website

207

Gear

44

FAQ

169

Database Listing

375

Staff

101

Discovery

2065111
Catch before they fall: a pose-guided attention framework for indoor safety
Falls can pose a serious health threat, especially for older people, often leading to fractures, head injuries, or long-term disability. This highlights the need for reliable and non-invasive detection systems. Existing solutions often suffer from limitations such as user discomfort with wearables, constrained coverage of fixed sensors, or environmental challenges in vision-based methods. This paper proposes a dual-stream attention-guided robust indoor fall detection framework to solve the problem. The approach combines a pose-guided stream with a video stream, in which the former captures high-level skeletal features through MediaPipe and simultaneously processes spatiotemporal dynamics from raw RGB frames using a 3D Convolutional Neural Network (CNN) with a temporal attention module. In order to enhance classification precision, features from each stream are adaptively fused in a block based on attention mechanisms, which improves the model’s interpretation of posture and movement semantics. Testing on the KFall dataset reveals that the proposed method achieves 98.71% accuracy, surpassing pre-existing benchmarks. Triggering a visual alert by displaying a red rectangle on the screen upon the occurrence of this event is a subsequent outcome of this work. An ablation study highlights the effectiveness of each component. Finally, the proposed work advances fall detection by combining pose estimation and attention-based deep learning to deliver an accurate, interpretable, and deployable solution.
Results of trawl counts for juvenile pink salmon in the Bering and Okhotsk Seas in 2024 and prospects for the returns and catch in the Karaginsky subzone and Okhotsk Sea in 2025
The key stage of juvenile pink salmon monitoring is the survey of their fall feeding in the sea that detects the year-class strength. The results of this survey are used for forecasting of the pink salmon returns to the Karaginsky fishing subzone in Bering Sea and to the Okhotsk Sea and the landing in the next year. The trawl surveys with two research vessels have conducted in recent years that allows to cover vast areas in a short time and to exclude repeated counts of the same fish on neighbor transects. In 2024, such trawl survey for pink salmon counting was conducted in the western Bering Sea and Okhotsk Sea that provided representative data on the juveniles abundance used for forecasting their returns and catches in the Karaginsky fishing subzone and Okhotsk Sea. The abundance of pink salmon in the western Bering Sea was estimated in 452∙106 ind. that was about a half of their numbers in previous even years (2018, 2020, 2022). Their return to the Karaginsky fishing subzone in the next year is expected as 72∙106 ind. at the lower limit of confidence interval, that provides the catch of 49∙103 t with the expected weight of spawners about 1 kg and 32 % escapement to the spawning grounds. In the Okhotsk Sea, the abundance of pink salmon was estimated in 1077∙106 ind. that was lower than in 2020 and 2022. Their expected return to the Okhotsk Sea in the next year is expected as 123∙106 ind. at the lower limit of confidence interval, that provides the catch of 100∙103 t with the weight of 1.3 kg and escapement of 35 %. Abundance in «northern» and «southern» regional complexes of local stocks is estimated for pink salmon in the Okhotsk Sea using cluster analysis with the expectation-maximization algorithm (EM clustering); the «northern» group prevailed with the ratio 64:36 %.