BEGIN:VCALENDAR
VERSION:2.0
PRODID:IEEE vTools.Events//EN
CALSCALE:GREGORIAN
BEGIN:VTIMEZONE
TZID:America/New_York
BEGIN:DAYLIGHT
DTSTART:20260308T030000
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
RRULE:FREQ=YEARLY;BYDAY=2SU;BYMONTH=3
TZNAME:EDT
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:20261101T010000
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=11
TZNAME:EST
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260401T143950Z
UID:C4CD70EF-3DA2-4167-B333-92E11600C5FC
DTSTART;TZID=America/New_York:20260327T130000
DTEND;TZID=America/New_York:20260327T143400
DESCRIPTION:Visitor Parking is next to the Engineering Building West Campus
 \n\nAbstract: In recent years\, 3D human activity recognition and tracking
  has become an important topic in human-computer interaction. To preserve 
 the privacy of users\, there is considerable interest in techniques withou
 t using a video camera. In this talk\, we first present RFID-Pose\, a visi
 on-assisted 3D human pose estimation system based on deep learning (DL). T
 he performance of DL models depends on the availability of sufficient high
 -quality radio frequency (RF) data\, which is more difficult and expensive
  to collect than other types of data. To overcome this obstacle\, in the s
 econd part of this talk\, we present generative AI approaches to generate 
 labeled synthetic RF data for multiple wireless sensing platforms\, such a
 s WiFi\, RFID\, and mmWave radar\, including a conditional Recurrent Gener
 ative Adversarial Network (R-GAN) approach and diffusion/latent diffusion 
 based approaches. Next\, we propose a novel framework that leverages laten
 t diffusion transformers to synthesize high quality RF data\, as well as a
  latent diffusion transformer with cross-attention conditioning to accurat
 ely infer missing joints in skeletal poses\, completing full 25-joint conf
 igurations from partial (i.e.\, 12-joint) inputs utilizing received RF sen
 sory data. Finally\, we present our recent work TF-Diff\, a novel training
 -free diffusion framework for cross-domain radio frequency (RF)-based huma
 n activity recognition (HAR) system\, which enables effective adaptation w
 ith minimal target-domain data.\n\nSpeaker(s): Shiwen Mao\, \n\nRoom: 221\
 , Bldg: Engineering Building \, 840 Polytechnic Ln\, Marietty\, Georgia\, 
 United States\, 30060
LOCATION:Room: 221\, Bldg: Engineering Building \, 840 Polytechnic Ln\, Mar
 ietty\, Georgia\, United States\, 30060
ORGANIZER:Bpage1@ieee.org
SEQUENCE:65
SUMMARY:Diffusion-enabled 3D Human Pose Tracking\, Data Augmentation\, Comp
 letion\, and Acceleration
URL;VALUE=URI:https://events.vtools.ieee.org/m/549541
X-ALT-DESC:Description: <br /><p>&nbsp\;</p>\n<p>Visitor Parking is next to
  the Engineering Building&nbsp\; West Campus&nbsp\;</p>\n<p>&nbsp\;</p>\n<
 p>Abstract: In recent years\, 3D human activity recognition and tracking h
 as become an important topic in human-computer interaction. To preserve th
 e privacy of users\, there is considerable interest in techniques without 
 using a video camera. In this talk\, we first present RFID-Pose\, a vision
 -assisted 3D human pose estimation system based on deep learning (DL). The
  performance of DL models depends on the availability of sufficient high-q
 uality radio frequency (RF) data\, which is more difficult and expensive t
 o collect than other types of data. To overcome this obstacle\, in the sec
 ond part of this talk\, we present generative AI approaches to generate la
 beled synthetic RF data for multiple wireless sensing platforms\, such as 
 WiFi\, RFID\, and mmWave radar\, including a conditional Recurrent Generat
 ive Adversarial Network (R-GAN) approach and diffusion/latent diffusion ba
 sed approaches. Next\, we propose a novel framework that leverages latent 
 diffusion transformers to synthesize high quality RF data\, as well as a l
 atent diffusion transformer with cross-attention conditioning to accuratel
 y infer missing joints in skeletal poses\, completing full 25-joint config
 urations from partial (i.e.\, 12-joint) inputs utilizing received RF senso
 ry data. Finally\, we present our recent work TF-Diff\, a novel training-f
 ree diffusion framework for cross-domain radio frequency (RF)-based human 
 activity recognition (HAR) system\, which enables effective adaptation wit
 h minimal target-domain data.&nbsp\;</p>
END:VEVENT
END:VCALENDAR