Upload 5 files

sd/prompt_helper/helper.py

@@ -0,0 +1,23 @@
+import onnx
+import onnxruntime as ort
+import numpy as np
+
+VIT_MODEL_DSV3_REPO = "SmilingWolf/wd-vit-tagger-v3"
+
+
+def preprocess_image(img):
+    bgr_img = np.array(img)[:, :, ::-1].copy()
+    size = max(bgr_img.shape[0:2])
+    pad_x = size - bgr_img.shape[1]
+    pad_y = size - bgr_img.shape[0]
+    pad_l = pad_x // 2
+    pad_t = pad_y // 2
+
+    #add paddings to squaring image
+    np.pad(bgr_img, ((pad_t, pad_y - pad_t), (pad_l, pad_x - pad_l), (0, 0)), mode="constant", constant_values=255)
+
+    #adaptive resize
+    interp = cv2.INTER_AREA if size > IMAGE_SIZE else cv2.INTER_LANCZOS4
+    bgr_img = cv2.resize(bgr_img, (IMAGE_SIZE, IMAGE_SIZE), interpolation=interp)
+
+    bgr_img = bgr_img.astype(np.float32)

sd/prompt_helper/model/config.json

@@ -0,0 +1,39 @@
+{
+  "architecture": "vit_base_patch16_224",
+  "num_classes": 10861,
+  "num_features": 768,
+  "global_pool": "avg",
+  "model_args": {
+    "img_size": 448,
+    "class_token": false,
+    "global_pool": "avg",
+    "fc_norm": false,
+    "act_layer": "gelu_tanh"
+  },
+  "pretrained_cfg": {
+    "custom_load": false,
+    "input_size": [
+      3,
+      448,
+      448
+    ],
+    "fixed_input_size": false,
+    "interpolation": "bicubic",
+    "crop_pct": 1.0,
+    "crop_mode": "center",
+    "mean": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "std": [
+      0.5,
+      0.5,
+      0.5
+    ],
+    "num_classes": 10861,
+    "pool_size": null,
+    "first_conv": null,
+    "classifier": null
+  }
+}

sd/prompt_helper/model/model.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:35f23693620b668f4d53fd3c62bf65e40af739bc52c7eb0fbc49258b58d065b6
+size 378536310

sd/prompt_helper/model/sw_jax_cv_config.json

@@ -0,0 +1,14 @@
+{
+    "image_size": 448,
+    "model_name": "vit_base",
+    "model_args": {
+        "patch_size": 16,
+        "num_classes": 10861,
+        "num_layers": 12,
+        "embed_dim": 768,
+        "mlp_dim": 3072,
+        "num_heads": 12,
+        "drop_path_rate": 0.1,
+        "layer_norm_eps": 1e-05
+    }
+}