upload y11 models and script

yolo-07-2025/eyes-detection-01-fp32.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:edaf66727e5e98a62a8d3602b46d8fdcc8d999c7cf64a7c8d139c13fd6a90f60
+size 10457446

yolo-07-2025/eyes-detection-01.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b63370cdce9dac61417c40e7fb0da8753b3347c0e9d6fe67554c7a4c7f2bc79b
+size 5272929

yolo-07-2025/hs-real-anime-y11n-320-fp16.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c6519f33a6032006867b96a0a8e9ca3069a900b9a2ba1a0670f7669c313d2d7b
+size 5279134

yolo-07-2025/hs-real-anime-y11n-320-fp32.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8874ef5b98a60fee1579a1724665168c8334d424ed2d8344026dccfdf354d26e
+size 10469501

yolo-07-2025/hs-real-anime-y11n-640-fp16.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e26ff81a9295755fac7568e8c1531e9757abf2bfbcac8f79c4526838e9e02572
+size 5342193

yolo-07-2025/hs-real-anime-y11n-640-fp32.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fecf08b62cf091df6b46222e163ac1721570d5f0a9cffa93591e5bb46f61bccf
+size 10595556

yolo-07-2025/hs-real-anime-y11s-320-fp16.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0c2aeb9d199a9bdf0facb23ee8c878235347eb25e134b8162638dbfc5329a772
+size 18946830

yolo-07-2025/hs-real-anime-y11s-320-fp32.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7980bf8f1732f16466bc6c5997db2c9685422045a021536697a0de7845940d1b
+size 37804611

yolo-07-2025/hs-real-anime-y11s-640-fp16.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a3f3e85afff182b7a4157e1fca37b61e63c7eee315a3f47a9112f2f9919c44c5
+size 19009889

yolo-07-2025/hs-real-anime-y11s-640-fp32.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bd2f67c628adb20ab2f1ffdd45c37015103f6fa6140e1a6a942dbbc3121a44a9
+size 37930666

yolo-07-2025/hs-real-y11n-320-fp16.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:baeb88f900efef2b1b1d996ca8dc9bf28514f8db0d7ce6b3e94e1ea09cd12b9e
+size 5279134

yolo-07-2025/hs-real-y11n-320-fp32.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:39fd6000eeb729bfd5489f897f38eac6dca67c1008b320a7ed032a250e28b444
+size 10469501

yolo-07-2025/hs-real-y11n-640-fp16.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c8439d90a684247cded62c1de4f081a16d6ddcf069c40fa2fc967a79ba99f09
+size 5342193

yolo-07-2025/hs-real-y11n-640-fp32.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b2b5adff81442762a93f78cfdad4b492471c92fa4e60fc6d553afffd0dc892d5
+size 10595556

yolo-07-2025/labelize-folder.py

@@ -0,0 +1,155 @@
+import os
+import cv2
+import numpy as np
+import onnxruntime as ort
+
+
+def letterbox(image, new_shape=(320, 320), color=(114, 114, 114)):
+    """
+    Resize image to a target size with unchanged aspect ratio using padding.
+    Returns the resized image, and scaling/padding info to map boxes back.
+    """
+    shape = image.shape[:2]  # current shape [height, width]
+    r = min(new_shape[0] / shape[0], new_shape[1] / shape[1])
+    ratio = (r, r)
+    new_unpad = (int(round(shape[1] * r)), int(round(shape[0] * r)))
+    dw, dh = new_shape[1] - new_unpad[0], new_shape[0] - new_unpad[1]
+    dw /= 2
+    dh /= 2
+
+    # resize image
+    resized = cv2.resize(image, new_unpad, interpolation=cv2.INTER_LINEAR)
+    # add border
+    top, bottom = int(round(dh - 0.1)), int(round(dh + 0.1))
+    left, right = int(round(dw - 0.1)), int(round(dw + 0.1))
+    padded = cv2.copyMakeBorder(resized, top, bottom, left, right, cv2.BORDER_CONSTANT, value=color)
+    return padded, ratio, (dw, dh)
+
+
+def preprocess(image, input_shape):
+    """
+    Preprocess image for model inference: letterbox, BGR to RGB, transpose, normalize.
+    """
+    img, ratio, (dw, dh) = letterbox(image, new_shape=input_shape)
+    img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
+    img = img.astype(np.float32) / 255.0
+    img = img.transpose(2, 0, 1)
+    img = np.expand_dims(img, 0)
+    return img, ratio, (dw, dh)
+
+
+def postprocess(prediction, image_shape, ratio, pad, score_threshold=0.2, iou_threshold=0.45):
+    """
+    Convert model output to bounding boxes, apply threshold and OpenCV NMS.
+    Returns list of (class_id, score, box) in original image coords.
+    """
+    pred = prediction[0]  # (4+nc, N)
+    nc = pred.shape[0] - 4
+    pred = pred.transpose(1, 0)  # (N, 4+nc)
+
+    # Extract boxes and scores
+    boxes = pred[:, :4]
+    scores_all = pred[:, 4:]
+    class_ids = np.argmax(scores_all, axis=1)
+    confidences = scores_all[np.arange(len(class_ids)), class_ids]
+
+    # Filter by confidence
+    mask = confidences > score_threshold
+    boxes = boxes[mask]
+    confidences = confidences[mask]
+    class_ids = class_ids[mask]
+    if boxes.size == 0:
+        return []
+
+    # Convert from [cx, cy, w, h] to [x1, y1, x2, y2]
+    dx, dy = pad
+    gain = ratio[0]
+    boxes_xyxy = np.zeros_like(boxes)
+    boxes_xyxy[:, 0] = boxes[:, 0] - boxes[:, 2] / 2  # x1
+    boxes_xyxy[:, 1] = boxes[:, 1] - boxes[:, 3] / 2  # y1
+    boxes_xyxy[:, 2] = boxes[:, 0] + boxes[:, 2] / 2  # x2
+    boxes_xyxy[:, 3] = boxes[:, 1] + boxes[:, 3] / 2  # y2
+    boxes_xyxy[:, [0, 2]] = (boxes_xyxy[:, [0, 2]] - dx) / gain
+    boxes_xyxy[:, [1, 3]] = (boxes_xyxy[:, [1, 3]] - dy) / gain
+
+    # Clip to image size
+    h0, w0 = image_shape
+    boxes_xyxy[:, [0, 2]] = boxes_xyxy[:, [0, 2]].clip(0, w0)
+    boxes_xyxy[:, [1, 3]] = boxes_xyxy[:, [1, 3]].clip(0, h0)
+
+    # Prepare for OpenCV NMSBoxes: convert to [x, y, w, h]
+    xywh = []
+    for x1, y1, x2, y2 in boxes_xyxy:
+        xywh.append([float(x1), float(y1), float(x2 - x1), float(y2 - y1)])
+
+    # Apply NMS
+    indices = cv2.dnn.NMSBoxes(xywh, confidences.tolist(), score_threshold, iou_threshold)
+    if len(indices) == 0:
+        return []
+    indices = indices.flatten()
+
+    results = []
+    for i in indices:
+        x1, y1, x2, y2 = boxes_xyxy[i]
+        cls = int(class_ids[i])
+        conf = float(confidences[i])
+        results.append((cls, conf, (x1, y1, x2, y2)))
+    return results
+
+
+def save_labels(labels, image_shape, save_path):
+    """
+    Save labels in YOLO format: class_id, x_center, y_center, width, height (normalized).
+    """
+    h, w = image_shape
+    with open(save_path, 'w') as f:
+        for cls, conf, (x1, y1, x2, y2) in labels:
+            xc = (x1 + x2) / 2 / w
+            yc = (y1 + y2) / 2 / h
+            bw = (x2 - x1) / w
+            bh = (y2 - y1) / h
+            f.write(f"{cls} {xc:.6f} {yc:.6f} {bw:.6f} {bh:.6f}\n")
+
+
+def labelize_folder(folder, session, display=False):
+    """
+    Process all images in a folder: run detection, save labels, optionally display images.
+    """
+    input_name = session.get_inputs()[0].name
+    H, W = session.get_inputs()[0].shape[2:]
+
+    for filename in os.listdir(folder):
+        if not (filename.lower().endswith('.jpg') or filename.lower().endswith('.png')):
+            continue
+        img_path = os.path.join(folder, filename)
+        image = cv2.imread(img_path)
+        if image is None:
+            continue
+
+        img_input, ratio, pad = preprocess(image, (H, W))
+        outputs = session.run(None, {input_name: img_input})[0]
+        labels = postprocess(outputs, image.shape[:2], ratio, pad)
+
+        txt_path = os.path.splitext(img_path)[0] + '.txt'
+        save_labels(labels, image.shape[:2], txt_path)
+
+        if display:
+            for cls, conf, (x1, y1, x2, y2) in labels:
+                cv2.rectangle(image, (int(x1), int(y1)), (int(x2), int(y2)), (0, 255, 0), 2)
+                cv2.putText(image, f"{cls}:{conf:.2f}", (int(x1), int(y1)-5), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0,255,0), 1)
+            cv2.imshow('Detections', image)
+            cv2.waitKey(0)
+            cv2.destroyAllWindows()
+
+
+if __name__ == '__main__':
+    # Folder with images to process
+    folder = './dataset'
+    # Path to ONNX model
+    model_path = './hs-real-y11n-640-fp32.onnx'
+
+    # Create ONNX Runtime session
+    sess = ort.InferenceSession(model_path)
+
+    # Run labelization
+    labelize_folder(folder, sess, display=False)

yolo-07-2025/labels.txt

@@ -0,0 +1,16 @@
+FEMALE_FACE
+MALE_FACE
+FEMALE_GENITALIA_COVERED
+FEMALE_GENITALIA_EXPOSED
+BUTTOCKS_COVERED
+BUTTOCKS_EXPOSED
+FEMALE_BREAST_COVERED
+FEMALE_BREAST_EXPOSED
+MALE_BREAST_EXPOSED
+ARMPITS_EXPOSED
+BELLY_EXPOSED
+MALE_GENITALIA_EXPOSED
+ANUS_EXPOSED
+FEET_COVERED
+FEET_EXPOSED
+EYE