Upload 4 files

README.md

@@ -22,14 +22,14 @@ model-index:
       name: Video Classification
     metrics:
     - type: accuracy
-      value: 0.9515
+      value: 0.9683
       name: Accuracy
     - type: f1
-      value: 0.9392
+      value: 0.9600
       name: Macro F1
 ---
 
-# Driver Behavior Detection Model (Epoch 2)
+# Driver Behavior Detection Model (Epoch 4)
 
 운전자 이상행동 감지를 위한 Video Swin Transformer 기반 모델입니다.
 
@@ -44,18 +44,18 @@ model-index:
 
 | Label | Class | F1-Score |
 |:-----:|-------|:--------:|
-| 0 | 정상 (Normal) | 0.93 |
-| 1 | 졸음운전 (Drowsy Driving) | 0.98 |
-| 2 | 물건찾기 (Reaching/Searching) | 0.90 |
-| 3 | 휴대폰 사용 (Phone Usage) | 0.88 |
+| 0 | 정상 (Normal) | 0.95 |
+| 1 | 졸음운전 (Drowsy Driving) | 0.99 |
+| 2 | 물건찾기 (Reaching/Searching) | 0.94 |
+| 3 | 휴대폰 사용 (Phone Usage) | 0.93 |
 | 4 | 운전자 폭행 (Driver Assault) | 1.00 |
 
-## Performance (Epoch 2)
+## Performance (Epoch 4)
 
 | Metric | Value |
 |--------|-------|
-| **Accuracy** | 95.15% |
-| **Macro F1** | 0.9392 |
+| **Accuracy** | 96.83% |
+| **Macro F1** | 0.9600 |
 | **Validation Samples** | 1,371,062 |
 
 ## Training Configuration
@@ -64,14 +64,14 @@ model-index:
 |-----------|-------|
 | Hardware | 2x NVIDIA RTX A6000 (48GB) |
 | Distributed | DDP (DistributedDataParallel) |
-| Batch Size | 32 (16 × 2 GPU) |
+| Batch Size | 32 (16 x 2 GPU) |
 | Gradient Accumulation | 4 |
 | Effective Batch | 128 |
 | Optimizer | AdamW (lr=1e-3, wd=0.05) |
 | Scheduler | OneCycleLR |
 | Mixed Precision | FP16 |
 | Loss | CrossEntropy + Label Smoothing (0.1) |
-| Regularization | Mixup (α=0.4), Dropout (0.3) |
+| Regularization | Mixup (a=0.4), Dropout (0.3) |
 
 ## Usage
 
@@ -98,7 +98,7 @@ with torch.no_grad():
 - **Total Samples (windows)**: 1,371,062
 - **Window Size**: 30 frames
 - **Stride**: 15 frames
-- **Resolution**: 224×224
+- **Resolution**: 224x224
 
 ## Augmentation (Training)
 
@@ -106,7 +106,7 @@ with torch.no_grad():
 - HorizontalFlip (p=0.5)
 - ColorJitter, HueSaturationValue
 - Temporal Augmentation (speed change, frame drop)
-- Mixup (α=0.4)
+- Mixup (a=0.4)
 - CoarseDropout
 
 ## License

config.json

@@ -11,9 +11,9 @@
     "layers": ["LayerNorm(768)", "Dropout(0.3)", "Linear(768, 5)"]
   },
   "training": {
-    "epoch": 2,
-    "accuracy": 0.9515,
-    "macro_f1": 0.9392,
+    "epoch": 4,
+    "accuracy": 0.9683,
+    "macro_f1": 0.9600,
     "batch_size": 32,
     "optimizer": "AdamW",
     "learning_rate": 1e-3,
@@ -23,10 +23,10 @@
     "augmentation": ["Mixup(0.4)", "RandomResizedCrop", "HorizontalFlip", "ColorJitter", "TemporalAugmentation"]
   },
   "performance": {
-    "정상": {"precision": 0.91, "recall": 0.95, "f1": 0.93},
-    "졸음운전": {"precision": 0.99, "recall": 0.97, "f1": 0.98},
-    "물건찾기": {"precision": 0.92, "recall": 0.88, "f1": 0.90},
-    "휴대폰 사용": {"precision": 0.84, "recall": 0.93, "f1": 0.88},
+    "정상": {"precision": 0.95, "recall": 0.94, "f1": 0.95},
+    "졸음운전": {"precision": 0.99, "recall": 0.98, "f1": 0.99},
+    "물건찾기": {"precision": 0.93, "recall": 0.95, "f1": 0.94},
+    "휴대폰 사용": {"precision": 0.92, "recall": 0.93, "f1": 0.93},
     "운전자 폭행": {"precision": 1.00, "recall": 1.00, "f1": 1.00}
   }
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ae9125be6e38460b5519ca5fc0bad96e952297b1858a95bd15ebaa7d0a772f3f
+oid sha256:346b59e7df4bf056e98b43f7e4f6a43f495516429899f7301679cea380fac98a
 size 126244047