Spaces:
Sleeping
Sleeping
File size: 3,432 Bytes
c66d0d6 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 |
import cv2
import gradio as gr
from ultralytics import YOLO
# ── Config ─────────────────────────────────────────────
MODEL_PATH = "yolov8n.pt" # modelo pre-entrenado (clase “person”)
CONF_THRES = 0.3 # confianza mínima detección
LINE_RATIO = 0.5 # posición de la línea virtual (50 % altura)
# ───────────────────────────────────────────────────────
model = YOLO(MODEL_PATH)
# estado global
memory = {} # {track_id: previous_cy}
in_count = 0
out_count = 0
def count_people(frame):
"""
Recibe un frame RGB (numpy) -> procesa -> devuelve frame RGB anotado y string.
Se llama de forma continua porque el input tiene `streaming=True`.
"""
global memory, in_count, out_count
if frame is None:
return None, ""
# ── paso 1: RGB ➜ BGR para OpenCV/YOLO ─────────────
frame_bgr = cv2.cvtColor(frame, cv2.COLOR_RGB2BGR)
h, w = frame_bgr.shape[:2]
line_y = int(h * LINE_RATIO)
# ── detección + tracking ───────────────────────────
results = model.track(
frame_bgr,
classes=[0], # solo “person”
conf=CONF_THRES,
persist=True,
verbose=False
)
annotated = frame_bgr.copy()
cv2.line(annotated, (0, line_y), (w, line_y), (0, 255, 255), 2)
if results:
for box in results[0].boxes:
x1, y1, x2, y2 = map(int, box.xyxy[0])
cx, cy = int((x1 + x2) / 2), int((y1 + y2) / 2)
tid = int(box.id[0]) if box.id is not None else -1
# cruces entrada / salida
prev_cy = memory.get(tid, cy)
if prev_cy < line_y <= cy: # cruzó de arriba → abajo (ENTRA)
in_count += 1
elif prev_cy > line_y >= cy: # abajo → arriba (SALE)
out_count += 1
memory[tid] = cy
# dibujitos
cv2.rectangle(annotated, (x1, y1), (x2, y2), (0, 255, 0), 1)
cv2.circle(annotated, (cx, cy), 3, (0, 0, 255), -1)
cv2.putText(annotated, str(tid), (x1, y1 - 5),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (255, 0, 0), 1)
total = in_count - out_count
label = f"In: {in_count} | Out: {out_count} | Ocupación: {total}"
# ── paso 2: BGR ➜ RGB para mostrar en Gradio ───────
annotated_rgb = cv2.cvtColor(annotated, cv2.COLOR_BGR2RGB)
return annotated_rgb, label
def reset_counts():
"""Callback para botón ‘Limpiar’."""
global memory, in_count, out_count
memory = {}
in_count = 0
out_count = 0
return None, ""
with gr.Blocks(title="Contador de personas (entrada única)") as demo:
gr.Markdown("# Contador de personas (entrada única)")
with gr.Row():
cam = gr.Image(sources=["webcam"], streaming=True, label="frame")
out_img = gr.Image(label="Video")
out_lbl = gr.Text(label="Contador")
btn_clear = gr.Button("Limpiar")
# wire-up
cam.stream(fn=count_people, outputs=[out_img, out_lbl])
btn_clear.click(fn=reset_counts, outputs=[out_img, out_lbl])
if __name__ == "__main__":
demo.launch()
|