feat: init

.gitattributes

@@ -0,0 +1,50 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bin.* filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+ 
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zstandard filter=lfs diff=lfs merge=lfs -text
+*.tfevents* filter=lfs diff=lfs merge=lfs -text
+*.db* filter=lfs diff=lfs merge=lfs -text
+*.ark* filter=lfs diff=lfs merge=lfs -text
+**/*ckpt*data* filter=lfs diff=lfs merge=lfs -text
+**/*ckpt*.meta filter=lfs diff=lfs merge=lfs -text
+**/*ckpt*.index filter=lfs diff=lfs merge=lfs -text
+ 
+ 
+*.gguf* filter=lfs diff=lfs merge=lfs -text
+*.ggml filter=lfs diff=lfs merge=lfs -text
+*.llamafile* filter=lfs diff=lfs merge=lfs -text
+*.pt2 filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text
+
+best.ckpt filter=lfs diff=lfs merge=lfs -text
+model.safetensors filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -0,0 +1,36 @@
+# Python
+__pycache__/
+*.py[oc]
+*.pyc
+*.pyo
+*.pyd
+
+# IDE
+.vscode/
+.idea/
+*.swp
+*.swo
+
+# OS
+.DS_Store
+Thumbs.db
+desktop.ini
+
+# Training outputs
+checkpoints/
+output/
+logs/
+tb_logs/
+wandb/
+runs/
+
+# Temporary files
+*.tmp
+*.temp
+
+# Secrets
+.env
+*.env
+secrets.json
+credentials.json
+api_keys.json

README.md

@@ -0,0 +1,35 @@
+# ExamInk-Seg
+
+基于 PyTorch Lightning 的墨迹二值分割模型。输入 RGB 图像，输出 1 通道墨迹掩码。
+
+## 模型信息
+
+- **架构**：U-Net++ (segmentation_models_pytorch) + ResNet50 编码器 (ImageNet 预训练)
+- **任务**：手写/试卷墨迹二值分割
+- **框架**：PyTorch Lightning
+- **输入**：3 通道 RGB，按 ImageNet 均值/方差归一化
+- **输出**：1 通道 logits 掩码；对 logits 取 sigmoid 得到概率
+
+## 训练概要
+
+- **Loss**：0.8 * Dice (from logits) + 0.2 * BCEWithLogits (pos_weight=2.0)
+- **指标**：IoU 与 F1（阈值 0.3 和 0.5）
+- **训练裁剪**：1024，随机裁剪 / 有墨迹优先裁剪，水平翻转
+- **优化器**：AdamW (lr 3e-5, weight_decay 1e-5)
+- **学习率调度**：ReduceLROnPlateau，监控 val_iou_05（mode=max, factor=0.5, patience=2, min_lr=1e-7）
+
+## 文件
+
+- `config.json`：模型结构与训练超参数
+- `model.safetensors`：导出权重
+- `best.ckpt`：PyTorch Lightning checkpoint
+
+## 推理说明
+
+- 归一化：mean=(0.485, 0.456, 0.406)，std=(0.229, 0.224, 0.225)
+- 对 logits 取 `sigmoid` 得到二值掩码概率
+- 验证/推理使用滑窗：tile=1024，stride=768，可选水平翻转 TTA
+
+## 许可证
+
+MIT

best.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8621c86f398fa2d4c298ffce66a683c25ba367ba3ea4449a81c2cf138e5c6241
+size 614284713

config.json

@@ -0,0 +1,69 @@
+{
+  "architectures": [
+    "InkEraserModel"
+  ],
+  "model_type": "unet_plus_plus",
+  "encoder_name": "resnet50",
+  "encoder_weights": "imagenet",
+  "in_channels": 3,
+  "classes": 1,
+  "decoder_attention_type": "scse",
+  "activation": "sigmoid",
+  "framework": "pytorch-lightning",
+  "task": "binary-segmentation",
+  "input_normalization": {
+    "mean": [
+      0.485,
+      0.456,
+      0.406
+    ],
+    "std": [
+      0.229,
+      0.224,
+      0.225
+    ],
+    "max_pixel_value": 255.0
+  },
+  "training_config": {
+    "optimizer": "AdamW",
+    "lr": 3e-05,
+    "weight_decay": 1e-05,
+    "loss": {
+      "dice_weight": 0.8,
+      "bce_weight": 0.2,
+      "bce_pos_weight": 2.0,
+      "dice_from_logits": true
+    },
+    "metrics": [
+      "iou_03",
+      "iou_05",
+      "f1_03",
+      "f1_05"
+    ],
+    "scheduler": {
+      "name": "ReduceLROnPlateau",
+      "mode": "max",
+      "monitor": "val_iou_05",
+      "factor": 0.5,
+      "patience": 2,
+      "min_lr": 1e-07
+    },
+    "train": {
+      "batch_size": 6,
+      "crop_size": 1024,
+      "accumulate_grad_batches": 4,
+      "precision": "bf16-mixed",
+      "max_epochs": 200,
+      "min_epochs": 5
+    }
+  },
+  "inference": {
+    "sliding_window": {
+      "tile": 1024,
+      "stride": 768,
+      "tta_hflip": true
+    }
+  },
+  "description": "ExamInk-Seg 二值墨迹分割模型（U-Net++ + ResNet50 编码器）。",
+  "license": "MIT"
+}

configuration.json

@@ -0,0 +1 @@
+{"framework":"pytorch-lightning","task":"binary-segmentation"}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f23fa08ab21b5acce191c29c3f96dced710ad36cd20a319ca4bdb79f9432c42f
+size 204810588