increase image editor zone

adjust example path

Add faint.mp4 file under Git LFS

Configure Git LFS to track .mp4 files

add a royality-free example video

use mid-hip to determine if a person is inside of alert zone but keep using bottom centor for velocity estimation

show skeleton

use mid-hip to detct if a person appear inside of alert zone instead of using bottom center which may lead to some false-alarm

Apply low-pass filtering on velocity estimator output

adjust motionless timer lower limit in ui

adjust ui laytout

Update README.md

Use Yolov11s pose to unify all models

fix readme

update readme

enabel zerogpu

show alert zone in red line

add alert zone define zone, and detc faint only in the zone

remove environmental classification

Add a demo on how to make ui letting user draw on a video preview

decrease gpu dur to 200 sec

increase zero gpu duration

improve annotation layout

Analysiss result readibility improvement

bugfix

velocity threshold +environmental classification result

update readme

add 2nd method based on velocity for motionlessness detection

update readme

apply rtdetr_r50vd_coco_o365 for exnvironmental classification

fix model id for rt-detr

add context awareness by environmental classification

parse vitpose output

do kpt post processing diy

bugfix to ViTPose processing error: Batch of images and boxes mismatch

add missing arg to image_processor call

add debug info

rescaled pose measue to be aligned with bbox scale

again, tuen para in is_lying_from_pose

tune parameter in is_lying_from_pose

undebug

debug with vitpose bypassed

back to vitpose small

yolov11s

pure yolov5

switch to yolov5

fix for zero gpu

fix for vitpose plus

add transformers

vitpose-small