Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[detection] model training iteration with extended multilingual dataset #1604

Open
9 tasks
Tracked by #1856
felixdittrich92 opened this issue May 22, 2024 · 1 comment
Open
9 tasks
Tracked by #1856

[detection] model training iteration with extended multilingual dataset #1604

felixdittrich92 opened this issue May 22, 2024 · 1 comment
Assignees
@felixdittrich92 @SiddhantBahuguna
Labels
benchmark This issue reports a benchmark framework: pytorch Related to PyTorch backend framework: tensorflow Related to TensorFlow backend topic: text detection Related to the task of text detection type: enhancement Improvement
Milestone
0.12.0

Comments

@felixdittrich92
Copy link
Contributor

felixdittrich92 commented May 22, 2024 •
edited
Loading

🚀 The feature

Retrain detection models from scratch with the updated augmentation pipeline & multilingual extended detection dataset

  • db_mobilenet_v3_large
  • db_resnet50
  • db_resnet34
  • linknet_resnet18
  • linknet_resnet34
  • linknet_resnet50
  • fast_tiny (can be ported to TF @felixdittrich92 )
  • fast_small (can be ported to TF @felixdittrich92)
  • fast_base (can be ported to TF @felixdittrich92)

args:

--rotation
--save_interval
--epochs 20

benchmarks: @felixdittrich92

goal:

  • Improved results on non latin text
  • Model robustness and improvements on edge cases like:

#1535
#1547
#1562
#1570
#1795
@felixdittrich92 felixdittrich92 added type: enhancement Improvement benchmark This issue reports a benchmark framework: pytorch Related to PyTorch backend framework: tensorflow Related to TensorFlow backend topic: text detection Related to the task of text detection labels May 22, 2024
This was referenced May 22, 2024
Closed
Closed
@felixdittrich92 felixdittrich92 added this to the 0.10.0 milestone Jun 6, 2024
@felixdittrich92 felixdittrich92 mentioned this issue Jun 6, 2024
Closed
@felixdittrich92 felixdittrich92 modified the milestones: 0.10.0, 0.11.0 Oct 10, 2024
@felixdittrich92 felixdittrich92 mentioned this issue Oct 11, 2024
Closed
3 tasks
This was referenced Dec 5, 2024
Closed
Closed
Closed
@felixdittrich92
Copy link
Contributor Author

Additional:

  • non latin text like chinese, cyrillic, thai, arabic -> detection improvements
  • dense "small" text instances
  • large text like newspaper headlines
  • additional elements like bullet points

@felixdittrich92 felixdittrich92 mentioned this issue Jan 28, 2025
Open
4 tasks
@felixdittrich92 felixdittrich92 modified the milestones: 0.11.0, 0.12.0 Jan 28, 2025
@felixdittrich92 felixdittrich92 changed the title [experimential] [detection] model training iteration with updated augmentation pipeline [detection] model training iteration with extended multilingual dataset Mar 5, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@felixdittrich92 felixdittrich92

@SiddhantBahuguna SiddhantBahuguna

Labels
benchmark This issue reports a benchmark framework: pytorch Related to PyTorch backend framework: tensorflow Related to TensorFlow backend topic: text detection Related to the task of text detection type: enhancement Improvement
Projects
None yet
Milestone
0.12.0
Development

No branches or pull requests
3 participants
@felixdittrich92 @SiddhantBahuguna @odulcy-mindee