Update README.md
Browse files
README.md
CHANGED
|
@@ -20,13 +20,24 @@ Annotations included manually sorting each of the 60 breeds into a catagory base
|
|
| 20 |
| Percent| 87% | 8% | 4% |
|
| 21 |
| Count | 20,731 | 1,975 | 988 |
|
| 22 |
|
| 23 |
-
|
| 24 |
***
|
| 25 |
# Training Prcedure
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 26 |
***
|
| 27 |
# Evaluation Resuls
|
| 28 |
### *Comprehensive Metrics*
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 29 |
### *Detailed Per-Class Breakdown*
|
|
|
|
| 30 |
### *Examples Of Classes*
|
| 31 |
### *Visualizations*
|
| 32 |
### *Performance Analysis*
|
|
|
|
| 20 |
| Percent| 87% | 8% | 4% |
|
| 21 |
| Count | 20,731 | 1,975 | 988 |
|
| 22 |
|
|
|
|
| 23 |
***
|
| 24 |
# Training Prcedure
|
| 25 |
+
* Trained on Google Collab using A100
|
| 26 |
+
* Limited to 200 epochs and 100 patience
|
| 27 |
+
* Ran for 73 epochs, best at 63
|
| 28 |
+
* 4.9 Hours training on ~24k images
|
| 29 |
+
* 10k base
|
| 30 |
+
* 14k augmented on exposure and blur
|
| 31 |
***
|
| 32 |
# Evaluation Resuls
|
| 33 |
### *Comprehensive Metrics*
|
| 34 |
+
| Metric | Goal | Result |
|
| 35 |
+
|--------|--------|----------|
|
| 36 |
+
|F1 Score | .75 < | .84 |
|
| 37 |
+
|Recall | .75 < | .96 |
|
| 38 |
+
|Precision | .80 < | 1.0 |
|
| 39 |
### *Detailed Per-Class Breakdown*
|
| 40 |
+
|
| 41 |
### *Examples Of Classes*
|
| 42 |
### *Visualizations*
|
| 43 |
### *Performance Analysis*
|