Upload RT-DETRv2 voucher classifier
Browse files
README.md
CHANGED
@@ -58,32 +58,21 @@ This model is a fine-tuned version of [PekingU/rtdetr_v2_r101vd](https://hugging
|
|
58 |
|
59 |
### Performance Metrics
|
60 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
61 |
**Final Evaluation Results:**
|
62 |
-
**Overall Detection Performance:**
|
63 |
-
- **mAP**: 0.0000
|
64 |
-
- **mAP@50**: 0.0000
|
65 |
-
- **mAP@75**: 0.0000
|
66 |
-
|
67 |
-
**Per-Class Average Precision:**
|
68 |
-
- **Digital invoices**: 0.0000 (needs improvement)
|
69 |
-
- **Fisico receipts**: 0.0000 (needs improvement)
|
70 |
-
- **Tesoreria receipts**: 0.0000 (needs improvement)
|
71 |
-
|
72 |
-
**Model Confidence:**
|
73 |
-
- **Digital invoices mean confidence**: 0.7044 (moderate)
|
74 |
-
- **Fisico receipts mean confidence**: 0.5937 (low)
|
75 |
-
- **Tesoreria receipts mean confidence**: 0.5683 (low)
|
76 |
-
|
77 |
-
**Performance by Object Size:**
|
78 |
-
- **Small objects**: 0.0000
|
79 |
-
- **Medium objects**: -1.0000
|
80 |
-
- **Large objects**: 0.0000
|
81 |
-
|
82 |
-
**Evaluation Dataset:**
|
83 |
-
- **Digital invoices**: 157 samples (28.5%)
|
84 |
-
- **Fisico receipts**: 261 samples (47.4%)
|
85 |
-
- **Tesoreria receipts**: 133 samples (24.1%)
|
86 |
-
- **Total evaluation samples**: 551
|
87 |
|
88 |
**Model Configuration:**
|
89 |
- **Base model**: PekingU/rtdetr_v2_r101vd
|
@@ -98,15 +87,15 @@ This model is a fine-tuned version of [PekingU/rtdetr_v2_r101vd](https://hugging
|
|
98 |
- **RAM**: 83.5 GB
|
99 |
- **GPU configuration**: A100 optimized
|
100 |
|
101 |
-
**Training Time**: 0.
|
102 |
|
103 |
**Training Summary:**
|
104 |
-
- **Final training loss**:
|
105 |
|
106 |
|
107 |
### MLflow Tracking
|
108 |
|
109 |
-
- **MLflow Run ID**:
|
110 |
- **MLflow Experiment**: RT-DETRv2_Voucher_Classification
|
111 |
|
112 |
|
|
|
58 |
|
59 |
### Performance Metrics
|
60 |
|
61 |
+
**Metric Definitions:**
|
62 |
+
|
63 |
+
- **mAP (mean Average Precision)**: Overall performance metric averaged across all classes and IoU thresholds (0.0-1.0, higher is better)
|
64 |
+
- **mAP@50**: mAP calculated at IoU threshold 0.5 - more lenient, measures if objects are found in roughly correct location
|
65 |
+
- **mAP@75**: mAP calculated at IoU threshold 0.75 - more strict, requires precise bounding box localization
|
66 |
+
- **IoU (Intersection over Union)**: Overlap between predicted and ground truth bounding boxes
|
67 |
+
|
68 |
+
**Performance Ranges:**
|
69 |
+
- 0.9+: Excellent
|
70 |
+
- 0.8-0.9: Very Good
|
71 |
+
- 0.7-0.8: Good
|
72 |
+
- 0.5-0.7: Fair
|
73 |
+
- <0.5: Poor (needs improvement)
|
74 |
+
|
75 |
**Final Evaluation Results:**
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
76 |
|
77 |
**Model Configuration:**
|
78 |
- **Base model**: PekingU/rtdetr_v2_r101vd
|
|
|
87 |
- **RAM**: 83.5 GB
|
88 |
- **GPU configuration**: A100 optimized
|
89 |
|
90 |
+
**Training Time**: 0.6 minutes
|
91 |
|
92 |
**Training Summary:**
|
93 |
+
- **Final training loss**: 1361.9480
|
94 |
|
95 |
|
96 |
### MLflow Tracking
|
97 |
|
98 |
+
- **MLflow Run ID**: 65eb62e7fd564f99981143809773def8
|
99 |
- **MLflow Experiment**: RT-DETRv2_Voucher_Classification
|
100 |
|
101 |
|
checkpoint-22/model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 306699044
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:41bcea9c8faca4547536af694dd5ad4a304153763de5c5d40387fb9d71bedb70
|
3 |
size 306699044
|
checkpoint-22/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 611580433
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:443919119e128db40e45a79d6a18a299f9ef64236143bc3b0a00dc2b02f890c4
|
3 |
size 611580433
|
checkpoint-22/training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5368
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ee47b52e482bd2dc4e70d1d92d8db4b33fc8cc8bb349a91709c9f357b574e2d1
|
3 |
size 5368
|
model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 306699044
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:41bcea9c8faca4547536af694dd5ad4a304153763de5c5d40387fb9d71bedb70
|
3 |
size 306699044
|
runs/Aug14_00-08-42_9db0f8c974d2/events.out.tfevents.1755130123.9db0f8c974d2.91164.0
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:98d25d95362299af66d8224db094bd64647151b9f37e98fed010630d17a98975
|
3 |
+
size 7402
|
training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5368
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ee47b52e482bd2dc4e70d1d92d8db4b33fc8cc8bb349a91709c9f357b574e2d1
|
3 |
size 5368
|