BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Chicago
X-LIC-LOCATION:America/Chicago
BEGIN:DAYLIGHT
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
TZNAME:CDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
TZNAME:CST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20230124T170804Z
LOCATION:C1-2-3
DTSTART;TZID=America/Chicago:20221116T083000
DTEND;TZID=America/Chicago:20221116T170000
UID:submissions.supercomputing.org_SC22_sess274_rpost107@linklings.com
SUMMARY:Debris Pose Estimation Using Deep Learning on FPGA
DESCRIPTION:Posters, Research Posters\n\nDebris Pose Estimation Using Deep
  Learning on FPGA\n\nHashimoto\n\nIt is difficult to implement a CNN for e
 dge processing in satellites, automobiles, and more, where machine resourc
 es and power are limited. FPGAs meet such constraints of machine resources
  and power associated with CNNs. FPGAs have low power consumption, but lim
 ited machine resources. Quantization Neural Networks have fewer parameters
  (bit depth) than CNNs and better estimation accuracy than BNNs.\n\nAlthou
 gh CNNs for regression problems are rarely implemented with FPGAs, our stu
 dy installed debris pose estimation on an FPGA using the latest edge techn
 ology such as quantization neural network. Pose estimations were run on a 
 workstation using 32bit floating-point precision and on an FPGA using 8bit
  int precision. The average errors were 4.98% and 5.38%, respectively. Thi
 s demonstrates that the regression problem can be transferred to an FPGA w
 ithout a significant loss of accuracy. The FPGA power efficiency is more t
 han 218k times that of a workstation implementation.\n\nRegistration Categ
 ory: Tech Program Reg Pass, Exhibits Reg Pass
END:VEVENT
END:VCALENDAR