Skip to content
VuFind
  • Login
    • English
    • اللغة العربية
Advanced
  • Comparison of different loss f...
  • Cite this
  • Text this
  • Email this
  • Print
  • Export Record
    • Export to RefWorks
    • Export to EndNoteWeb
    • Export to EndNote
  • Save to List
  • Permanent link
Comparison of different loss functions (%). Verify the optimal combination weight of Dice Loss and BCE Loss.

Comparison of different loss functions (%). Verify the optimal combination weight of Dice Loss and BCE Loss.

<p>Comparison of different loss functions (%). Verify the optimal combination weight of Dice Loss and BCE Loss.</p>

Saved in:
Bibliographic Details
Main Author: Kerang Cao (21602045) (author)
Other Authors: Miao Zhao (527763) (author), Minghui Geng (21602048) (author), Shuai Zheng (482786) (author), Hoekyung Jung (21602051) (author)
Published: 2025
Subjects:
Biotechnology
Space Science
Biological Sciences not elsewhere classified
Mathematical Sciences not elsewhere classified
Information Systems not elsewhere classified
unlike conventional methods
reducing model size
reduce computational overhead
large model sizes
experimental results reveal
effectively capture multi
combine binary cross
cardiac function assessments
cardiac function assessment
address gradient issues
proposed model introduces
dice loss functions
enhance feature selectivity
simple attention module
paper introduces
dice coefficient
attention module
attention mechanism
xlink ">
video data
stabilize training
scale features
optimized unet
minimize redundancy
lightweight simam
imprecise segmentation
expert annotations
efficient solution
dynamic dataset
automated diagnosis
also provide
Tags: Add Tag
No Tags, Be the first to tag this record!
  • Holdings
  • Description
  • Comments
  • Similar Items
  • Staff View
Be the first to leave a comment!
You must be logged in first

Similar Items

  • Model comparison. A comparison of DA-UNet++ and other models in terms of parameter quantity and segmentation performance was made, where the abscissa represents the model’s parameter quantity, and the ordinate represents the model’s Dice similarity coefficient on the EchoNet-Dynamic dataset.
    by: Kerang Cao (21602045)
    Published: (2025)
  • The role of each module for the overall framework (%). To evaluate the necessity of each module, the segmentation results improved with the addition of each module, and the complete method achieved the best results.
    by: Kerang Cao (21602045)
    Published: (2025)
  • Comparison of DSC with different cutting scales (%). Evaluating the impact of different scales of UNet++ on the model can reduce the network scale to lower computational and storage overhead while ensuring the accuracy of the model.
    by: Kerang Cao (21602045)
    Published: (2025)
  • Analysis of the segmentation performance of different attention mechanisms (%). By comparing the segmentation performance of different attention mechanisms, it is highlighted that SCBAM exhibits better segmentation performance.
    by: Kerang Cao (21602045)
    Published: (2025)
  • ESV of the left ventricle, The white line represents the ground truth, and the red line represents the prediction of the current model (i.e., the model labeled in the figure).
    by: Kerang Cao (21602045)
    Published: (2025)

Find More

  • Browse the Catalog
  • Browse Alphabetically
  • Explore Channels
  • Course Reserves
  • New Items
Cannot write session to /tmp/vufind_sessions/sess_7hd56h8lmoteekhqj5lis507n4