scorevision: push artifact

miner.py

@@ -1,24 +1,24 @@
-"""Petrol-station detection miner for SN44 (TurboVision).
+"""TurboVision crime-detection miner.
 
-Element: manak0_Detect-petrol-station-1-0  (4 classes: hose, pump, board, canopy)
-Backbone: YOLO11s, exported with NMS baked into the ONNX graph,
-          weights converted to FP16. Output shape: [1, 300, 6] =
-          (x1, y1, x2, y2, conf, cls_id) in letterboxed input coordinates.
+YOLO11s @ 1280x1280, 6-class detection (balaclava, bat, glove, graffiti, hoodie,
+spray paint), ONNX with end-to-end NMS baked in.
+
+Output of weights.onnx: [1, 300, 6] = x1, y1, x2, y2, conf, cls (post-NMS).
 
 Inference pipeline:
-  1. Letterbox to 1280x1280, BGR->RGB, /255, NCHW float16.
-  2. Single ORT CUDA pass + horizontal-flip TTA (boxes merged via per-class
-     hard NMS). Multi-scale TTA was net-negative on val so omitted here.
-  3. Per-class confidence thresholds tuned from the bench:
-       hose=0.55, pump=0.55, board=0.50, canopy=0.55.
-  4. Reverse letterbox; clip to image bounds; emit BoundingBox.
-
-Local val composite_onchain (multi-TTA): 0.6460 — leader median is 0.6276.
+  1) Primary forward pass on the full image.
+  2) Hflip TTA: forward on horizontally-flipped image, transform boxes back.
+  3) Per-class hard-NMS to merge primary + flip outputs.
+  4) Cross-class IoU dedup (suppresses same physical object getting two class labels).
+  5) Consensus-confidence boost: when both views agree on a cluster, take max score.
+  6) Sanity filter (min size, aspect ratio).
+
+Class taxonomy (must match the validator manifest's `objects` list for this element):
+  0 balaclava   1 bat   2 glove   3 graffiti   4 hoodie   5 spray paint
 """
-from __future__ import annotations
 
-import math
 from pathlib import Path
+import math
 
 import cv2
 import numpy as np
@@ -42,191 +42,410 @@ class TVFrameResult(BaseModel):
     keypoints: list[tuple[int, int]]
 
 
-CLASS_NAMES = ["petrol hose", "petrol pump", "price board", "roof canopy"]
-PER_CLASS_CONF = {0: 0.55, 1: 0.55, 2: 0.50, 3: 0.55}
-GLOBAL_CONF = min(PER_CLASS_CONF.values())  # filter floor before per-class
-NMS_IOU = 0.50
-WEIGHTS_FILENAME = "best_fp16.onnx"
-
-
-def _letterbox(image: ndarray, new_shape: tuple[int, int],
-               color: tuple[int, int, int] = (114, 114, 114)
-               ) -> tuple[ndarray, float, tuple[float, float]]:
-    h, w = image.shape[:2]
-    new_w, new_h = new_shape
-    ratio = min(new_w / w, new_h / h)
-    rw, rh = int(round(w * ratio)), int(round(h * ratio))
-    if (rw, rh) != (w, h):
-        interp = cv2.INTER_CUBIC if ratio > 1.0 else cv2.INTER_LINEAR
-        image = cv2.resize(image, (rw, rh), interpolation=interp)
-    dw = (new_w - rw) / 2.0
-    dh = (new_h - rh) / 2.0
-    left = int(round(dw - 0.1))
-    right = int(round(dw + 0.1))
-    top = int(round(dh - 0.1))
-    bottom = int(round(dh + 0.1))
-    padded = cv2.copyMakeBorder(image, top, bottom, left, right,
-                                cv2.BORDER_CONSTANT, value=color)
-    return padded, ratio, (dw, dh)
-
-
-def _hard_nms(boxes: np.ndarray, scores: np.ndarray, iou_t: float) -> np.ndarray:
-    if len(boxes) == 0:
-        return np.array([], dtype=np.intp)
-    order = np.argsort(scores)[::-1]
-    keep: list[int] = []
-    while len(order):
-        i = int(order[0])
-        keep.append(i)
-        if len(order) == 1:
-            break
-        rest = order[1:]
-        xx1 = np.maximum(boxes[i, 0], boxes[rest, 0])
-        yy1 = np.maximum(boxes[i, 1], boxes[rest, 1])
-        xx2 = np.minimum(boxes[i, 2], boxes[rest, 2])
-        yy2 = np.minimum(boxes[i, 3], boxes[rest, 3])
-        inter = np.maximum(0, xx2 - xx1) * np.maximum(0, yy2 - yy1)
-        area_i = (boxes[i, 2] - boxes[i, 0]) * (boxes[i, 3] - boxes[i, 1])
-        area_r = (boxes[rest, 2] - boxes[rest, 0]) * (boxes[rest, 3] - boxes[rest, 1])
-        iou = inter / (area_i + area_r - inter + 1e-9)
-        order = rest[iou <= iou_t]
-    return np.array(keep, dtype=np.intp)
-
-
 class Miner:
-    """SN44 chute entrypoint. Required:
-        - class named `Miner`
-        - method `predict_batch(batch_images, offset, n_keypoints)`
-        - file at the root of the HF repo as `miner.py`
-    """
-
     def __init__(self, path_hf_repo: Path) -> None:
-        model_path = path_hf_repo / WEIGHTS_FILENAME
-        if not model_path.exists():
-            # Defensive: try alternative names that earlier exports used
-            for alt in ("petrol.onnx", "weights.onnx", "best.onnx"):
-                if (path_hf_repo / alt).exists():
-                    model_path = path_hf_repo / alt
-                    break
-
-        print(f"[miner] ORT version: {ort.__version__}")
+        model_path = path_hf_repo / "weights.onnx"
+
+        # Validator manifest order (from spec.json `objects`):
+        #   0=balaclava 1=hoodie 2=glove 3=bat 4="spray paint" 5=graffiti
+        # v5 weights.onnx was trained with this exact order, so cls_remap is identity.
+        cn_path = model_path.with_name("class_names.txt")
+        if cn_path.is_file():
+            self.class_names = [
+                ln.strip()
+                for ln in cn_path.read_text(encoding="utf-8").splitlines()
+                if ln.strip() and not ln.strip().startswith("#")
+            ]
+        else:
+            self.class_names = ["balaclava", "hoodie", "glove", "bat", "spray paint", "graffiti"]
+        self.cls_remap = np.arange(len(self.class_names), dtype=np.int32)
+
+        print("ORT version:", ort.__version__)
         try:
             ort.preload_dlls()
-        except Exception:
-            pass
-        print(f"[miner] available providers: {ort.get_available_providers()}")
+            print("✅ onnxruntime.preload_dlls() success")
+        except Exception as e:
+            print(f"⚠️ preload_dlls failed: {e}")
+        print("ORT available providers BEFORE session:", ort.get_available_providers())
 
-        sess_opts = ort.SessionOptions()
-        sess_opts.graph_optimization_level = ort.GraphOptimizationLevel.ORT_ENABLE_ALL
+        sess_options = ort.SessionOptions()
+        sess_options.graph_optimization_level = ort.GraphOptimizationLevel.ORT_ENABLE_ALL
 
         try:
             self.session = ort.InferenceSession(
                 str(model_path),
-                sess_options=sess_opts,
+                sess_options=sess_options,
                 providers=["CUDAExecutionProvider", "CPUExecutionProvider"],
             )
-            print(f"[miner] ✅ session created (preferred CUDA)")
+            print("✅ Created ORT session with preferred CUDA provider list")
         except Exception as e:
-            print(f"[miner] ⚠️ CUDA session failed, falling back to CPU: {e}")
+            print(f"⚠️ CUDA session creation failed, falling back to CPU: {e}")
             self.session = ort.InferenceSession(
                 str(model_path),
-                sess_options=sess_opts,
+                sess_options=sess_options,
                 providers=["CPUExecutionProvider"],
             )
-
-        print(f"[miner] active providers: {self.session.get_providers()}")
-
-        self.input_name = self.session.get_inputs()[0].name
-        ishape = self.session.get_inputs()[0].shape
-        self.in_h = ishape[2] if isinstance(ishape[2], int) and ishape[2] > 0 else 1280
-        self.in_w = ishape[3] if isinstance(ishape[3], int) and ishape[3] > 0 else 1280
-        ttype = self.session.get_inputs()[0].type
-        self.dtype = np.float16 if "float16" in ttype else np.float32
-        print(f"[miner] input ({self.in_h}x{self.in_w}, dtype={self.dtype.__name__})")
-        print(f"[miner] weights: {model_path.name} ({model_path.stat().st_size/1e6:.2f} MB)")
-        print(f"[miner] per-class conf: {PER_CLASS_CONF}")
+        print("ORT session providers:", self.session.get_providers())
+
+        inp = self.session.get_inputs()[0]
+        self.input_name = inp.name
+        self.output_names = [o.name for o in self.session.get_outputs()]
+        self.input_shape = inp.shape
+        self.input_dtype = np.float16 if "float16" in inp.type else np.float32
+
+        self.input_height = self._safe_dim(self.input_shape[2], default=1280)
+        self.input_width = self._safe_dim(self.input_shape[3], default=1280)
+
+        # Tuning matched to alfred's deployed model — bias toward precision to dodge
+        # the false_positive pillar penalty (validator weights FP heavily on this element).
+        self.conf_thres = 0.50
+        self.iou_thres = 0.4
+        self.cross_iou_thresh = 0.7
+        self.max_det = 200
+        self.use_tta = True
+
+        # Sanity filter — reject obviously bad boxes
+        self.min_box_area = 14 * 14
+        self.min_side = 8
+        self.max_aspect_ratio = 8.0
+        self.max_box_area_ratio = 0.95
+
+        print(f"✅ ONNX loaded: {model_path}")
+        print(f"✅ providers: {self.session.get_providers()}")
+        print(f"✅ input: name={self.input_name}, shape={self.input_shape}, dtype={self.input_dtype}")
+        print(f"✅ classes: {self.class_names}")
+        print(f"✅ config: conf={self.conf_thres}, iou={self.iou_thres}, "
+              f"cross_iou={self.cross_iou_thresh}, TTA={self.use_tta}")
 
     def __repr__(self) -> str:
         return (
-            f"PetrolStationMiner(providers={self.session.get_providers()}, "
-            f"in={self.in_h}x{self.in_w}, dtype={self.dtype.__name__})"
+            f"ONNXRuntime(session={type(self.session).__name__}, "
+            f"providers={self.session.get_providers()})"
         )
 
-    def _run_pass(self, image: ndarray) -> list[BoundingBox]:
-        """One ONNX forward pass. Returns BoundingBox in original-image coords."""
-        h, w = image.shape[:2]
-        padded, ratio, (dw, dh) = _letterbox(image, (self.in_w, self.in_h))
-        rgb = cv2.cvtColor(padded, cv2.COLOR_BGR2RGB).astype(np.float32) / 255.0
-        x = np.transpose(rgb, (2, 0, 1))[None].astype(self.dtype, copy=False)
-        x = np.ascontiguousarray(x)
-        out = self.session.run(None, {self.input_name: x})[0]
-        if out.ndim == 3:
-            out = out[0]
-        out = out.astype(np.float32, copy=False)
-
-        valid = out[:, 4] > 0
-        if not valid.any():
-            return []
-        out = out[valid]
+    @staticmethod
+    def _safe_dim(value, default: int) -> int:
+        return value if isinstance(value, int) and value > 0 else default
 
-        boxes = out[:, :4].copy()
-        boxes[:, [0, 2]] -= dw
-        boxes[:, [1, 3]] -= dh
-        boxes /= ratio
+    def _letterbox(
+        self,
+        image: ndarray,
+        new_shape: tuple[int, int],
+        color=(114, 114, 114),
+    ) -> tuple[ndarray, float, tuple[float, float]]:
+        h, w = image.shape[:2]
+        new_w, new_h = new_shape
+        ratio = min(new_w / w, new_h / h)
+        resized_w = int(round(w * ratio))
+        resized_h = int(round(h * ratio))
+        if (resized_w, resized_h) != (w, h):
+            interp = cv2.INTER_CUBIC if ratio > 1.0 else cv2.INTER_LINEAR
+            image = cv2.resize(image, (resized_w, resized_h), interpolation=interp)
+        dw = (new_w - resized_w) / 2.0
+        dh = (new_h - resized_h) / 2.0
+        left = int(round(dw - 0.1))
+        right = int(round(dw + 0.1))
+        top = int(round(dh - 0.1))
+        bottom = int(round(dh + 0.1))
+        padded = cv2.copyMakeBorder(
+            image, top, bottom, left, right,
+            borderType=cv2.BORDER_CONSTANT, value=color,
+        )
+        return padded, ratio, (dw, dh)
+
+    def _preprocess(self, image: ndarray):
+        orig_h, orig_w = image.shape[:2]
+        img, ratio, pad = self._letterbox(image, (self.input_width, self.input_height))
+        img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
+        img = img.astype(self.input_dtype) / 255.0
+        img = np.transpose(img, (2, 0, 1))[None, ...]
+        img = np.ascontiguousarray(img)
+        return img, ratio, pad, (orig_w, orig_h)
+
+    @staticmethod
+    def _clip_boxes(boxes: np.ndarray, image_size: tuple[int, int]) -> np.ndarray:
+        w, h = image_size
         boxes[:, 0] = np.clip(boxes[:, 0], 0, w - 1)
         boxes[:, 1] = np.clip(boxes[:, 1], 0, h - 1)
         boxes[:, 2] = np.clip(boxes[:, 2], 0, w - 1)
         boxes[:, 3] = np.clip(boxes[:, 3], 0, h - 1)
-        scores = out[:, 4]
-        cls_ids = out[:, 5].astype(np.int32)
+        return boxes
 
-        results: list[BoundingBox] = []
-        for (x1, y1, x2, y2), c, k in zip(boxes, scores, cls_ids):
-            cls = int(k)
-            thr = PER_CLASS_CONF.get(cls, GLOBAL_CONF)
-            if c < thr or x2 <= x1 or y2 <= y1:
+    def _filter_sane_boxes(
+        self,
+        boxes: np.ndarray,
+        scores: np.ndarray,
+        cls_ids: np.ndarray,
+        orig_size: tuple[int, int],
+    ):
+        if len(boxes) == 0:
+            return boxes, scores, cls_ids
+        orig_w, orig_h = orig_size
+        image_area = float(orig_w * orig_h)
+        keep = []
+        for i, box in enumerate(boxes):
+            x1, y1, x2, y2 = box.tolist()
+            bw = x2 - x1
+            bh = y2 - y1
+            if bw <= 0 or bh <= 0:
                 continue
-            results.append(BoundingBox(
-                x1=int(math.floor(x1)),
-                y1=int(math.floor(y1)),
-                x2=int(math.ceil(x2)),
-                y2=int(math.ceil(y2)),
-                cls_id=cls,
-                conf=float(c),
-            ))
-        return results
+            if bw < self.min_side or bh < self.min_side:
+                continue
+            area = bw * bh
+            if area < self.min_box_area:
+                continue
+            if area > self.max_box_area_ratio * image_area:
+                continue
+            ar = max(bw / max(bh, 1e-6), bh / max(bw, 1e-6))
+            if ar > self.max_aspect_ratio:
+                continue
+            keep.append(i)
+        if not keep:
+            return (
+                np.empty((0, 4), dtype=np.float32),
+                np.empty((0,), dtype=np.float32),
+                np.empty((0,), dtype=np.int32),
+            )
+        k = np.array(keep, dtype=np.intp)
+        return boxes[k], scores[k], cls_ids[k]
+
+    @staticmethod
+    def _hard_nms(
+        boxes: np.ndarray,
+        scores: np.ndarray,
+        iou_thresh: float,
+    ) -> np.ndarray:
+        N = len(boxes)
+        if N == 0:
+            return np.array([], dtype=np.intp)
+        boxes = np.asarray(boxes, dtype=np.float32)
+        scores = np.asarray(scores, dtype=np.float32)
+        order = np.argsort(scores)[::-1]
+        keep: list[int] = []
+        suppressed = np.zeros(N, dtype=bool)
+        for i in range(N):
+            idx = order[i]
+            if suppressed[idx]:
+                continue
+            keep.append(int(idx))
+            bi = boxes[idx]
+            for k in range(i + 1, N):
+                jdx = order[k]
+                if suppressed[jdx]:
+                    continue
+                bj = boxes[jdx]
+                xx1 = max(bi[0], bj[0])
+                yy1 = max(bi[1], bj[1])
+                xx2 = min(bi[2], bj[2])
+                yy2 = min(bi[3], bj[3])
+                inter = max(0.0, xx2 - xx1) * max(0.0, yy2 - yy1)
+                area_i = (bi[2] - bi[0]) * (bi[3] - bi[1])
+                area_j = (bj[2] - bj[0]) * (bj[3] - bj[1])
+                iou = inter / (area_i + area_j - inter + 1e-7)
+                if iou > iou_thresh:
+                    suppressed[jdx] = True
+        return np.array(keep, dtype=np.intp)
+
+    def _per_class_hard_nms(
+        self,
+        boxes: np.ndarray,
+        scores: np.ndarray,
+        cls_ids: np.ndarray,
+        iou_thresh: float,
+    ) -> np.ndarray:
+        if len(boxes) == 0:
+            return np.array([], dtype=np.intp)
+        all_keep: list[int] = []
+        for c in np.unique(cls_ids):
+            mask = cls_ids == c
+            indices = np.where(mask)[0]
+            keep = self._hard_nms(boxes[mask], scores[mask], iou_thresh)
+            all_keep.extend(indices[keep].tolist())
+        all_keep.sort()
+        return np.array(all_keep, dtype=np.intp)
+
+    @staticmethod
+    def _cross_class_dedup(
+        boxes: np.ndarray,
+        scores: np.ndarray,
+        cls_ids: np.ndarray,
+        iou_thresh: float,
+    ) -> tuple[np.ndarray, np.ndarray, np.ndarray]:
+        n = len(boxes)
+        if n <= 1:
+            return boxes, scores, cls_ids
+        boxes = np.asarray(boxes, dtype=np.float32)
+        scores = np.asarray(scores, dtype=np.float32)
+        cls_ids = np.asarray(cls_ids, dtype=np.int32)
+        areas = np.maximum(0.0, boxes[:, 2] - boxes[:, 0]) * np.maximum(
+            0.0, boxes[:, 3] - boxes[:, 1]
+        )
+        # Keep larger boxes first, then higher score.
+        order = np.lexsort((-scores, -areas))
+        suppressed = np.zeros(n, dtype=bool)
+        keep: list[int] = []
+        for i in order:
+            if suppressed[i]:
+                continue
+            keep.append(int(i))
+            bi = boxes[i]
+            xx1 = np.maximum(bi[0], boxes[:, 0])
+            yy1 = np.maximum(bi[1], boxes[:, 1])
+            xx2 = np.minimum(bi[2], boxes[:, 2])
+            yy2 = np.minimum(bi[3], boxes[:, 3])
+            inter = np.maximum(0.0, xx2 - xx1) * np.maximum(0.0, yy2 - yy1)
+            area_i = max(1e-7, float((bi[2] - bi[0]) * (bi[3] - bi[1])))
+            union = area_i + areas - inter + 1e-7
+            iou = inter / union
+            dup = iou > iou_thresh
+            dup[i] = False
+            suppressed |= dup
+        keep_idx = np.array(keep, dtype=np.intp)
+        return boxes[keep_idx], scores[keep_idx], cls_ids[keep_idx]
+
+    @staticmethod
+    def _max_score_per_cluster(
+        coords: np.ndarray,
+        scores: np.ndarray,
+        keep_indices: np.ndarray,
+        iou_thresh: float,
+    ) -> np.ndarray:
+        n_keep = len(keep_indices)
+        if n_keep == 0:
+            return np.array([], dtype=np.float32)
+        coords = np.asarray(coords, dtype=np.float32)
+        scores = np.asarray(scores, dtype=np.float32)
+        out = np.empty(n_keep, dtype=np.float32)
+        for i in range(n_keep):
+            idx = keep_indices[i]
+            bi = coords[idx]
+            xx1 = np.maximum(bi[0], coords[:, 0])
+            yy1 = np.maximum(bi[1], coords[:, 1])
+            xx2 = np.minimum(bi[2], coords[:, 2])
+            yy2 = np.minimum(bi[3], coords[:, 3])
+            inter = np.maximum(0.0, xx2 - xx1) * np.maximum(0.0, yy2 - yy1)
+            area_i = (bi[2] - bi[0]) * (bi[3] - bi[1])
+            areas_j = (coords[:, 2] - coords[:, 0]) * (coords[:, 3] - coords[:, 1])
+            iou = inter / (area_i + areas_j - inter + 1e-7)
+            in_cluster = iou >= iou_thresh
+            out[i] = float(np.max(scores[in_cluster]))
+        return out
+
+    def _decode_raw_dets(
+        self,
+        preds: np.ndarray,
+        ratio: float,
+        pad: tuple[float, float],
+        orig_size: tuple[int, int],
+    ) -> tuple[np.ndarray, np.ndarray, np.ndarray]:
+        """Decode end2end NMS output and return (boxes, scores, cls_ids)
+        in original image coordinates, after conf-threshold + remap + letterbox-reverse + sanity."""
+        if preds.ndim == 3 and preds.shape[0] == 1:
+            preds = preds[0]
+        if preds.ndim != 2 or preds.shape[1] < 6:
+            raise ValueError(f"Unexpected ONNX output shape: {preds.shape}")
+
+        boxes = preds[:, :4].astype(np.float32)
+        scores = preds[:, 4].astype(np.float32)
+        cls_ids = preds[:, 5].astype(np.int32)
+
+        valid = (cls_ids >= 0) & (cls_ids < len(self.cls_remap))
+        boxes, scores, cls_ids = boxes[valid], scores[valid], cls_ids[valid]
+        cls_ids = self.cls_remap[cls_ids]
+
+        keep = scores >= self.conf_thres
+        boxes = boxes[keep]
+        scores = scores[keep]
+        cls_ids = cls_ids[keep]
+        if len(boxes) == 0:
+            return (
+                np.empty((0, 4), dtype=np.float32),
+                np.empty((0,), dtype=np.float32),
+                np.empty((0,), dtype=np.int32),
+            )
+
+        pad_w, pad_h = pad
+        orig_w, orig_h = orig_size
+        boxes[:, [0, 2]] -= pad_w
+        boxes[:, [1, 3]] -= pad_h
+        boxes /= ratio
+        boxes = self._clip_boxes(boxes, (orig_w, orig_h))
+
+        boxes, scores, cls_ids = self._filter_sane_boxes(boxes, scores, cls_ids, orig_size)
+        return boxes, scores, cls_ids
+
+    def _forward(
+        self, image: np.ndarray
+    ) -> tuple[np.ndarray, np.ndarray, np.ndarray]:
+        x, ratio, pad, orig_size = self._preprocess(image)
+        out = self.session.run(self.output_names, {self.input_name: x})[0]
+        return self._decode_raw_dets(out, ratio, pad, orig_size)
+
+    def _predict_single(self, image: np.ndarray) -> list[BoundingBox]:
+        boxes, scores, cls_ids = self._forward(image)
+        if len(boxes) == 0:
+            return []
+        return self._build_results(boxes, scores, cls_ids)
+
+    def _predict_tta(self, image: np.ndarray) -> list[BoundingBox]:
+        """Hflip TTA: merge primary + flipped via per-class hard-NMS,
+        then cross-class dedup, with consensus-confidence boost."""
+        ow = image.shape[1]
+        b1, s1, c1 = self._forward(image)
 
-    def _predict_with_hflip(self, image: ndarray) -> list[BoundingBox]:
-        """Original + horizontal flip. Per-class hard NMS, IoU=0.50."""
-        a = self._run_pass(image)
         flipped = cv2.flip(image, 1)
-        b = self._run_pass(flipped)
-        w = image.shape[1]
-        b_unflipped = [
-            BoundingBox(
-                x1=w - bb.x2, y1=bb.y1, x2=w - bb.x1, y2=bb.y2,
-                cls_id=bb.cls_id, conf=bb.conf,
-            )
-            for bb in b
-        ]
-        all_boxes = a + b_unflipped
-        if not all_boxes:
+        b2, s2, c2 = self._forward(flipped)
+        if len(b2):
+            x1f = ow - b2[:, 2]
+            x2f = ow - b2[:, 0]
+            b2 = np.stack([x1f, b2[:, 1], x2f, b2[:, 3]], axis=1)
+
+        if len(b1) == 0 and len(b2) == 0:
+            return []
+
+        boxes = np.concatenate([b1, b2], axis=0) if len(b2) else b1
+        scores = np.concatenate([s1, s2], axis=0) if len(b2) else s1
+        cls_ids = np.concatenate([c1, c2], axis=0) if len(b2) else c1
+
+        keep = self._per_class_hard_nms(boxes, scores, cls_ids, self.iou_thres)
+        if len(keep) == 0:
             return []
+        keep = keep[: self.max_det]
 
-        by_cls: dict[int, list[BoundingBox]] = {}
-        for bb in all_boxes:
-            by_cls.setdefault(bb.cls_id, []).append(bb)
+        # Consensus-confidence boost: cluster by IoU and take max score.
+        boosted = self._max_score_per_cluster(boxes, scores, keep, self.iou_thres)
 
-        merged: list[BoundingBox] = []
-        for cls_id, lst in by_cls.items():
-            coords = np.array(
-                [[bb.x1, bb.y1, bb.x2, bb.y2] for bb in lst], dtype=np.float32
+        boxes = boxes[keep]
+        cls_ids = cls_ids[keep]
+        scores = boosted
+
+        boxes, scores, cls_ids = self._cross_class_dedup(
+            boxes, scores, cls_ids, self.cross_iou_thresh
+        )
+        if len(boxes) == 0:
+            return []
+
+        return self._build_results(boxes, scores, cls_ids)
+
+    def _build_results(
+        self, boxes: np.ndarray, scores: np.ndarray, cls_ids: np.ndarray
+    ) -> list[BoundingBox]:
+        results: list[BoundingBox] = []
+        for box, conf, cls_id in zip(boxes, scores, cls_ids):
+            x1, y1, x2, y2 = box.tolist()
+            if x2 <= x1 or y2 <= y1:
+                continue
+            results.append(
+                BoundingBox(
+                    x1=int(math.floor(x1)),
+                    y1=int(math.floor(y1)),
+                    x2=int(math.ceil(x2)),
+                    y2=int(math.ceil(y2)),
+                    cls_id=int(cls_id),
+                    conf=float(conf),
+                )
             )
-            scores = np.array([bb.conf for bb in lst], dtype=np.float32)
-            keep = _hard_nms(coords, scores, NMS_IOU)
-            for i in keep:
-                merged.append(lst[int(i)])
-        return merged
+        return results
 
     def predict_batch(
         self,
@@ -234,55 +453,32 @@ class Miner:
         offset: int,
         n_keypoints: int,
     ) -> list[TVFrameResult]:
-        """Detection-only element (no keypoints) — return n_keypoints zeros
-        per frame to keep the schema stable across challenge types."""
-        n_kp = max(0, int(n_keypoints))
         results: list[TVFrameResult] = []
-        for i, image in enumerate(batch_images):
-            frame_idx = offset + i
+        for frame_number_in_batch, image in enumerate(batch_images):
+            if image is None or not isinstance(image, np.ndarray) or image.ndim != 3:
+                results.append(
+                    TVFrameResult(
+                        frame_id=offset + frame_number_in_batch,
+                        boxes=[],
+                        keypoints=[(0, 0) for _ in range(max(0, int(n_keypoints)))],
+                    )
+                )
+                continue
+            if image.dtype != np.uint8:
+                image = image.astype(np.uint8)
             try:
-                if image is None or not isinstance(image, np.ndarray) \
-                        or image.ndim != 3 or image.shape[2] != 3:
-                    raise ValueError(f"bad image at frame {frame_idx}: {type(image)}")
-                if image.dtype != np.uint8:
-                    image = image.astype(np.uint8)
-                boxes = self._predict_with_hflip(image)
+                if self.use_tta:
+                    boxes = self._predict_tta(image)
+                else:
+                    boxes = self._predict_single(image)
             except Exception as e:
-                print(f"[miner] ⚠️ frame {frame_idx} failed: {e}")
+                print(f"⚠️ Inference failed for frame {offset + frame_number_in_batch}: {e}")
                 boxes = []
-            results.append(TVFrameResult(
-                frame_id=frame_idx,
-                boxes=boxes,
-                keypoints=[(0, 0) for _ in range(n_kp)],
-            ))
+            results.append(
+                TVFrameResult(
+                    frame_id=offset + frame_number_in_batch,
+                    boxes=boxes,
+                    keypoints=[(0, 0) for _ in range(max(0, int(n_keypoints)))],
+                )
+            )
         return results
-
-
-def main() -> None:
-    """Local smoke test: load miner from cwd, run on argv images or a blank."""
-    import sys
-
-    repo = Path(__file__).parent
-    miner = Miner(repo)
-    print(repr(miner))
-
-    images: list[np.ndarray] = []
-    if len(sys.argv) > 1:
-        for p in sys.argv[1:]:
-            img = cv2.imread(p)
-            if img is None:
-                raise ValueError(f"cannot read {p}")
-            images.append(img)
-    else:
-        images = [np.zeros((720, 1280, 3), dtype=np.uint8)]
-
-    results = miner.predict_batch(images, offset=0, n_keypoints=0)
-    for r in results:
-        print(f"frame {r.frame_id}: {len(r.boxes)} boxes")
-        for b in r.boxes:
-            name = CLASS_NAMES[b.cls_id] if 0 <= b.cls_id < len(CLASS_NAMES) else b.cls_id
-            print(f"  {name:12s} conf={b.conf:.3f} ({b.x1},{b.y1},{b.x2},{b.y2})")
-
-
-if __name__ == "__main__":
-    main()