encoder를 dino v2 공식 모델로 변경

encoder.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ab6d6bebf6eb8a3aa75071ff89c6f4e4f2877d325ac192428fe16254ffcf2396
-size 1213332927
+oid sha256:2fabcbe2e231b7b2035588329ec6df427c08baa8723eb31546c52738888013ca
+size 1218043280

model_info.json

@@ -1,112 +0,0 @@
-{
-  "model_architecture": {
-    "backbone": "vit_large_patch14_dinov2.lvd142m",
-    "feature_dim": 1024,
-    "total_parameters": 303252502,
-    "trainable_parameters": 24598,
-    "freeze_backbone": true
-  },
-  "input_specification": {
-    "image_size": [
-      224,
-      224
-    ],
-    "channels": 3,
-    "pixel_range": [
-      0.0,
-      1.0
-    ],
-    "normalization": {
-      "mean": [
-        0.485,
-        0.456,
-        0.406
-      ],
-      "std": [
-        0.229,
-        0.224,
-        0.225
-      ],
-      "description": "ImageNet normalization for DINOv2"
-    },
-    "input_format": "RGB",
-    "tensor_layout": "NCHW"
-  },
-  "output_specification": {
-    "heads": {
-      "scene": {
-        "num_classes": 6,
-        "output_type": "logits",
-        "activation": "softmax",
-        "classes": [
-          16000001,
-          16000002,
-          16000006,
-          16000008,
-          16000009,
-          16000011
-        ]
-      },
-      "concept": {
-        "num_classes": 3,
-        "output_type": "logits",
-        "activation": "softmax",
-        "classes": [
-          17000001,
-          17000002,
-          17000003
-        ]
-      },
-      "object": {
-        "num_classes": 13,
-        "output_type": "logits",
-        "activation": "softmax",
-        "classes": [
-          18000001,
-          18000002,
-          18000004,
-          18000005,
-          18000006,
-          18000007,
-          18000008,
-          18000009,
-          18000010,
-          18000012,
-          18000014,
-          18000016,
-          "unclassified"
-        ]
-      }
-    }
-  },
-  "class_mappings": {
-    "scene": {
-      "0": 16000001,
-      "1": 16000002,
-      "2": 16000006,
-      "3": 16000008,
-      "4": 16000009,
-      "5": 16000011
-    },
-    "concept": {
-      "0": 17000001,
-      "1": 17000002,
-      "2": 17000003
-    },
-    "object": {
-      "0": 18000001,
-      "1": 18000002,
-      "2": 18000004,
-      "3": 18000005,
-      "4": 18000006,
-      "5": 18000007,
-      "6": 18000008,
-      "7": 18000009,
-      "8": 18000010,
-      "9": 18000012,
-      "10": 18000014,
-      "11": 18000016,
-      "12": "unclassified"
-    }
-  }
-}

preprocessor_config.json

@@ -0,0 +1,27 @@
+{
+  "crop_size": {
+    "height": 224,
+    "width": 224
+  },
+  "do_center_crop": true,
+  "do_convert_rgb": true,
+  "do_normalize": true,
+  "do_rescale": true,
+  "do_resize": true,
+  "image_mean": [
+    0.485,
+    0.456,
+    0.406
+  ],
+  "image_processor_type": "BitImageProcessor",
+  "image_std": [
+    0.229,
+    0.224,
+    0.225
+  ],
+  "resample": 3,
+  "rescale_factor": 0.00392156862745098,
+  "size": {
+    "shortest_edge": 256
+  }
+}