Update README.md
#1
by Rjavenger - opened
README.md
CHANGED
|
@@ -11,4 +11,20 @@ tags:
|
|
| 11 |
- epoch:1
|
| 12 |
---
|
| 13 |
|
| 14 |
-
### RF-DETR with step learning rate scheduling and optimized hyperparameters
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
- epoch:1
|
| 12 |
---
|
| 13 |
|
| 14 |
+
### RF-DETR with step learning rate scheduling and optimized hyperparameters
|
| 15 |
+
|
| 16 |
+
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.
|
| 17 |
+
|
| 18 |
+
|
| 19 |
+
## Total Loss over epochs
|
| 20 |
+
|
| 21 |
+
|
| 22 |
+
## Per-class mAP@50:95
|
| 23 |
+
|
| 24 |
+
|
| 25 |
+
## Per-class Precision and Recall (Last Epoch)
|
| 26 |
+
|
| 27 |
+
|
| 28 |
+
## COCO mAP vs Epochs
|
| 29 |
+
|
| 30 |
+
|