BEGIN:VCALENDAR
VERSION:2.0
PRODID:IEEE vTools.Events//EN
CALSCALE:GREGORIAN
BEGIN:VTIMEZONE
TZID:Mexico/General
BEGIN:STANDARD
DTSTART:20221030T010000
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
TZNAME:CST
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260401T232319Z
UID:628B8AE3-D88F-4E3D-A4B9-90DD3A7CB4FB
DTSTART;TZID=Mexico/General:20260325T100000
DTEND;TZID=Mexico/General:20260325T110000
DESCRIPTION:In recent years\, 3D human activity recognition and tracking ha
 s become an important topic in human-computer interaction. To preserve the
  privacy of users\, there is considerable interest in techniques without u
 sing 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-qu
 ality radio frequency (RF) data\, which is more difficult and expensive to
  collect than other types of data. To overcome this obstacle\, in the seco
 nd part of the talk\, we present generative AI approaches to generate labe
 led synthetic RF data for multiple wireless sensing platforms\, such as Wi
 Fi\, RFID\, and mmWave radar\, including a conditional Recurrent Generativ
 e Adversarial Network (R-GAN) approach and diffusion/latent diffusion base
 d approaches. Next\, we propose a novel framework that leverages latent di
 ffusion transformers to synthesize high quality RF data\, as well as a lat
 ent diffusion transformer with cross-attention conditioning to accurately 
 infer missing joints in skeletal poses\, completing full 25-joint configur
 ations from partial (i.e.\, 12-joint) inputs utilizing received RF sensory
  data. Finally\, we present our recent work TF-Diff\, a novel training-fre
 e diffusion framework for cross-domain radio frequency (RF)-based human ac
 tivity recognition (HAR) system\, which enables effective adaptation with 
 minimal target-domain data.\n\nCo-sponsored by: Instituto Tecnológico de 
 Cuautla (TecNM-Cuautla)\n\nSpeaker(s): Shiwen Mao\n\nRoom: Emiliano Zapata
 \, Bldg: H\, Libramiento Cuautla-Oaxaca S/N\, Colonia  Juan Morales\, Cuau
 tla\, Morelos\, Mexico\, 62745\, Virtual: https://events.vtools.ieee.org/m
 /542574
LOCATION:Room: Emiliano Zapata\, Bldg: H\, Libramiento Cuautla-Oaxaca S/N\,
  Colonia  Juan Morales\, Cuautla\, Morelos\, Mexico\, 62745\, Virtual: htt
 ps://events.vtools.ieee.org/m/542574
ORGANIZER:jvazquezbu@hotmail.com
SEQUENCE:181
SUMMARY:Diffusion-enabled 3D Human Pose Tracking\, Data Augmentation\, Comp
 letion\, and Acceleration
URL;VALUE=URI:https://events.vtools.ieee.org/m/542574
X-ALT-DESC:Description: <br /><p>In recent years\, 3D human activity recogn
 ition and tracking has become an important topic in human-computer interac
 tion. To preserve the 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 dee
 p learning (DL). The performance of DL models depends on the availability 
 of sufficient high-quality radio frequency (RF) data\, which is more diffi
 cult and expensive to collect than other types of data. To overcome this o
 bstacle\, in the second part of the talk\, we present generative AI approa
 ches to generate labeled synthetic RF data for multiple wireless sensing p
 latforms\, such as WiFi\, RFID\, and mmWave radar\, including a conditiona
 l Recurrent Generative Adversarial Network (R-GAN) approach and diffusion/
 latent diffusion based approaches. Next\, we propose a novel framework tha
 t leverages latent diffusion transformers to synthesize high quality RF da
 ta\, as well as a latent diffusion transformer with cross-attention condit
 ioning to accurately infer missing joints in skeletal poses\, completing f
 ull 25-joint configurations from partial (i.e.\, 12-joint) inputs utilizin
 g received RF sensory data. Finally\, we present our recent work TF-Diff\,
  a novel training-free diffusion framework for cross-domain radio frequenc
 y (RF)-based human activity recognition (HAR) system\, which enables effec
 tive adaptation with minimal target-domain data.&nbsp\;</p>
END:VEVENT
END:VCALENDAR