Downloads · 30 days
0
step_scheduler.rf-detr-nano
Subh775/step_scheduler.rf-detr-nano
step_scheduler.rf-detr-nano is a object detection model from Subh775. Use it when you need objects located in an image. It is set up for adapter-transformers. The card lists the license as apache-2.0.
We fine-tuned RF-DETR using a step learning rate scheduler on a custom dataset. Within two epochs, the model achieved a +3.7 increase in mAP@50:95, with balanced improvement in classification and localization losses.…
Access
Public
Updated Dec 9, 2025
Repo size
2.3 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.pth2.3 GB · 100%
At a glance
- Task
- Object Detection
- Library
- adapter-transformers
- License
- apache-2.0
- Access
- Public
- Created
- Nov 26, 2025
- Updated
- Dec 9, 2025
- SHA
- 09fd3efa
Model card
From the Hugging Face model README
RF-DETR with step learning rate scheduling and optimized hyperparameters
We fine-tuned RF-DETR using a step learning rate scheduler on a custom dataset. Within two epochs, the model achieved a +3.7 increase in mAP@50:95, with balanced improvement in classification and localization losses. EMA weights consistently outperformed standard parameters, indicating stable convergence. Per-class analysis shows strong performance on well-represented categories like two-wheelers and trucks, while smaller or visually ambiguous classes such as minibuses remain challenging, suggesting future improvements via data balancing.
Total Loss over epochs
Per-class mAP@50:95
Per-class Precision and Recall (Last Epoch)
COCO mAP vs Epochs
- Task
- Object Detection
- Library
- adapter-transformers
- License
- apache-2.0
- Languages
- en
- Created
- Nov 26, 2025
- Updated
- Dec 9, 2025