Upload 9 files

README.md

@@ -0,0 +1,13 @@
+# Example Element Repo
+
+This is a minimal Hugging Face-style Element repo.
+
+## Run
+
+```bash
+uv run python entrypoint.py < sample_input.json
+```
+
+## Output
+
+The entrypoint runs the YOLOv8 model in `best.pt` against the input frames and returns Detect output.

best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:482f0782bc283651ff8365612c01097b4ab4ff3974d103ab03fd3b4b166456ac
+size 6249194

element.yaml

@@ -0,0 +1,3 @@
+version: 0.1.0
+element_type: Detect
+entrypoint: entrypoint.py

entrypoint.py

@@ -0,0 +1,82 @@
+import json
+import sys
+from io import BytesIO
+from pathlib import Path
+from typing import Any, Dict, List
+
+import requests
+from PIL import Image
+from ultralytics import YOLO
+
+
+def load_image(frame: str, base_dir: Path) -> Image.Image:
+    if frame.startswith("http://") or frame.startswith("https://"):
+        response = requests.get(frame, timeout=30)
+        response.raise_for_status()
+        return Image.open(BytesIO(response.content)).convert("RGB")
+
+    path = Path(frame)
+    if not path.is_absolute():
+        path = (Path.cwd() / path).resolve()
+        if not path.exists():
+            candidate = (base_dir / frame).resolve()
+            if candidate.exists():
+                path = candidate
+    return Image.open(path).convert("RGB")
+
+
+def detect(model: YOLO, image: Image.Image, frame_idx: int) -> List[Dict[str, Any]]:
+    results = model(image)
+    detections: List[Dict[str, Any]] = []
+    result = results[0]
+    names = result.names or model.names
+
+    for det_idx, box in enumerate(result.boxes):
+        xyxy = box.xyxy[0].tolist()
+        class_id = int(box.cls[0].item())
+        detections.append(
+            {
+                "frame_idx": frame_idx,
+                "class": names.get(class_id, str(class_id)),
+                "bbox": [float(x) for x in xyxy],
+                "score": float(box.conf[0].item()),
+                "track_id": f"f{frame_idx}-d{det_idx}",
+            }
+        )
+
+    return detections
+
+
+def main() -> None:
+    raw = sys.stdin.read().strip()
+    if not raw:
+        print("[]")
+        return
+    try:
+        payload = json.loads(raw)
+    except json.JSONDecodeError:
+        print("[]")
+        return
+
+    base_dir = Path(__file__).resolve().parent
+    model_path = base_dir / "best.pt"
+    if not model_path.exists():
+        print("[]")
+        return
+
+    model = YOLO(str(model_path))
+    frames = payload.get("frames", [])
+    output: List[Dict[str, Any]] = []
+
+    for frame_idx, frame in enumerate(frames):
+        try:
+            image = load_image(frame, base_dir)
+        except Exception:
+            continue
+        output.extend(detect(model, image, frame_idx))
+
+    print(json.dumps(output))
+
+
+if __name__ == "__main__":
+    main()

pyproject.toml

@@ -0,0 +1,8 @@
+[project]
+name = "vehicle-detect-element"
+version = "0.1.0"
+requires-python = ">=3.11"
+dependencies = [
+  "requests>=2.31.0",
+  "ultralytics>=8.0.0",
+]

sample_input.json

@@ -0,0 +1,6 @@
+{
+  "clip_id": "sha256:example",
+  "fps": 5,
+  "frames": ["example.jpg"],
+  "ts_start_ms": 1234567890
+}

sample_output.json

@@ -0,0 +1,62 @@
+[
+  {
+    "frame_idx": 0,
+    "class": "car",
+    "bbox": [
+      272.696044921875,
+      238.8153839111328,
+      304.9505615234375,
+      284.86334228515625
+    ],
+    "score": 0.7260516881942749,
+    "track_id": "f0-d0"
+  },
+  {
+    "frame_idx": 0,
+    "class": "car",
+    "bbox": [
+      317.4508056640625,
+      249.2779083251953,
+      351.03350830078125,
+      295.40264892578125
+    ],
+    "score": 0.6707994937896729,
+    "track_id": "f0-d1"
+  },
+  {
+    "frame_idx": 0,
+    "class": "car",
+    "bbox": [
+      404.1763610839844,
+      279.63201904296875,
+      470.1379089355469,
+      345.353759765625
+    ],
+    "score": 0.3774946928024292,
+    "track_id": "f0-d2"
+  },
+  {
+    "frame_idx": 0,
+    "class": "car",
+    "bbox": [
+      360.1339416503906,
+      231.34637451171875,
+      396.8412170410156,
+      269.111328125
+    ],
+    "score": 0.33625954389572144,
+    "track_id": "f0-d3"
+  },
+  {
+    "frame_idx": 0,
+    "class": "bus",
+    "bbox": [
+      337.4912414550781,
+      277.13092041015625,
+      395.4140930175781,
+      351.0
+    ],
+    "score": 0.3119567632675171,
+    "track_id": "f0-d4"
+  }
+]