Add face detection and presence tracking
Day 81 - Major upgrade! 🦊👀 NEW FEATURES: - Face detection using face-detection-retail-0004 on Myriad X - /presence endpoint - am I there? face count, last seen time - /face endpoint - detailed detection boxes and confidence - Background detection loop (every 0.5s) - Presence timeout after 30s without face Now Vixy can SEE when Foxy sits down! 💜 Technical: - Uses blobconverter for model download - MobileNetDetectionNetwork for on-device inference - Thread-safe presence state tracking - Added requirements.txt
This commit is contained in:
Alex Kazaiev
271
oak_service.py
271
oak_service.py
@@ -1,91 +1,270 @@
|
||||
#!/usr/bin/env python3
|
||||
"""
|
||||
OAK-D Vision Service for Vixy's Head
|
||||
FastAPI service exposing OAK-D camera capabilities
|
||||
FastAPI service with face detection and presence tracking
|
||||
|
||||
Day 74 - Built by Vixy! 🦊
|
||||
Day 81 - Added face detection + presence! Now I can SEE you! 👀💜
|
||||
"""
|
||||
|
||||
import io
|
||||
import asyncio
|
||||
import time
|
||||
import threading
|
||||
from contextlib import asynccontextmanager
|
||||
from fastapi import FastAPI, HTTPException
|
||||
from fastapi.responses import Response, JSONResponse
|
||||
from fastapi.responses import Response
|
||||
import depthai as dai
|
||||
import blobconverter
|
||||
import cv2
|
||||
import numpy as np
|
||||
|
||||
# Global device reference
|
||||
# ============== Configuration ==============
|
||||
FACE_DETECTION_MODEL = "face-detection-retail-0004"
|
||||
DETECTION_THRESHOLD = 0.5
|
||||
PRESENCE_TIMEOUT = 30.0 # seconds without face = not present
|
||||
DETECTION_INTERVAL = 0.5 # how often to check for faces
|
||||
|
||||
# ============== Global State ==============
|
||||
oak_device = None
|
||||
pipeline = None
|
||||
queue = None
|
||||
rgb_queue = None
|
||||
detection_queue = None
|
||||
detection_thread = None
|
||||
running = False
|
||||
|
||||
# Presence tracking state
|
||||
presence_state = {
|
||||
"present": False,
|
||||
"face_count": 0,
|
||||
"last_seen": None,
|
||||
"last_detection": None,
|
||||
"detections": [], # Current face bounding boxes
|
||||
"confidence": 0.0,
|
||||
}
|
||||
|
||||
|
||||
def init_oak():
|
||||
"""Initialize OAK-D camera with basic RGB pipeline."""
|
||||
global oak_device, pipeline, queue
|
||||
"""Initialize OAK-D with face detection pipeline."""
|
||||
global oak_device, pipeline, rgb_queue, detection_queue
|
||||
|
||||
try:
|
||||
oak_device = dai.Device()
|
||||
pipeline = dai.Pipeline(oak_device)
|
||||
cam = pipeline.create(dai.node.Camera).build(dai.CameraBoardSocket.CAM_A)
|
||||
queue = cam.requestFullResolutionOutput().createOutputQueue()
|
||||
pipeline.start()
|
||||
# Create pipeline
|
||||
pipeline = dai.Pipeline()
|
||||
|
||||
# RGB Camera
|
||||
cam_rgb = pipeline.create(dai.node.ColorCamera)
|
||||
cam_rgb.setPreviewSize(300, 300) # NN input size
|
||||
cam_rgb.setInterleaved(False)
|
||||
cam_rgb.setFps(10) # Lower FPS for efficiency
|
||||
|
||||
# Also get full resolution for snapshots
|
||||
cam_rgb.setResolution(dai.ColorCameraProperties.SensorResolution.THE_1080_P)
|
||||
|
||||
# Face detection neural network
|
||||
face_nn = pipeline.create(dai.node.MobileNetDetectionNetwork)
|
||||
face_nn.setConfidenceThreshold(DETECTION_THRESHOLD)
|
||||
face_nn.setBlobPath(blobconverter.from_zoo(
|
||||
name=FACE_DETECTION_MODEL,
|
||||
shaves=6,
|
||||
zoo_type="depthai"
|
||||
))
|
||||
face_nn.setNumInferenceThreads(2)
|
||||
face_nn.input.setBlocking(False)
|
||||
|
||||
# Link camera to NN
|
||||
cam_rgb.preview.link(face_nn.input)
|
||||
|
||||
# Output queues
|
||||
xout_rgb = pipeline.create(dai.node.XLinkOut)
|
||||
xout_rgb.setStreamName("rgb")
|
||||
cam_rgb.video.link(xout_rgb.input) # Full resolution for snapshots
|
||||
|
||||
xout_nn = pipeline.create(dai.node.XLinkOut)
|
||||
xout_nn.setStreamName("detections")
|
||||
face_nn.out.link(xout_nn.input)
|
||||
|
||||
# Start device
|
||||
oak_device = dai.Device(pipeline)
|
||||
rgb_queue = oak_device.getOutputQueue("rgb", maxSize=1, blocking=False)
|
||||
detection_queue = oak_device.getOutputQueue("detections", maxSize=1, blocking=False)
|
||||
|
||||
print("✅ OAK-D initialized with face detection!")
|
||||
return True
|
||||
|
||||
except Exception as e:
|
||||
print(f"Failed to initialize OAK-D: {e}")
|
||||
print(f"❌ Failed to initialize OAK-D: {e}")
|
||||
return False
|
||||
|
||||
|
||||
def cleanup_oak():
|
||||
"""Cleanup OAK-D resources."""
|
||||
global oak_device, pipeline, queue
|
||||
if pipeline:
|
||||
global oak_device, pipeline, rgb_queue, detection_queue, running
|
||||
running = False
|
||||
|
||||
if oak_device:
|
||||
try:
|
||||
pipeline.stop()
|
||||
oak_device.close()
|
||||
except:
|
||||
pass
|
||||
|
||||
oak_device = None
|
||||
pipeline = None
|
||||
queue = None
|
||||
rgb_queue = None
|
||||
detection_queue = None
|
||||
|
||||
|
||||
def detection_loop():
|
||||
"""Background thread that continuously checks for faces."""
|
||||
global running, presence_state, detection_queue
|
||||
|
||||
print("🔍 Face detection loop started")
|
||||
|
||||
while running:
|
||||
try:
|
||||
if detection_queue is None:
|
||||
time.sleep(1)
|
||||
continue
|
||||
|
||||
# Get detection results (non-blocking)
|
||||
in_nn = detection_queue.tryGet()
|
||||
|
||||
if in_nn is not None:
|
||||
detections = in_nn.detections
|
||||
now = time.time()
|
||||
|
||||
face_count = len(detections)
|
||||
|
||||
# Update presence state
|
||||
presence_state["last_detection"] = now
|
||||
presence_state["face_count"] = face_count
|
||||
|
||||
if face_count > 0:
|
||||
presence_state["present"] = True
|
||||
presence_state["last_seen"] = now
|
||||
presence_state["confidence"] = max(d.confidence for d in detections)
|
||||
presence_state["detections"] = [
|
||||
{
|
||||
"xmin": d.xmin,
|
||||
"ymin": d.ymin,
|
||||
"xmax": d.xmax,
|
||||
"ymax": d.ymax,
|
||||
"confidence": d.confidence
|
||||
}
|
||||
for d in detections
|
||||
]
|
||||
else:
|
||||
presence_state["detections"] = []
|
||||
presence_state["confidence"] = 0.0
|
||||
|
||||
# Check timeout
|
||||
if presence_state["last_seen"]:
|
||||
elapsed = now - presence_state["last_seen"]
|
||||
if elapsed > PRESENCE_TIMEOUT:
|
||||
presence_state["present"] = False
|
||||
|
||||
time.sleep(DETECTION_INTERVAL)
|
||||
|
||||
except Exception as e:
|
||||
print(f"Detection loop error: {e}")
|
||||
time.sleep(1)
|
||||
|
||||
print("🛑 Face detection loop stopped")
|
||||
|
||||
|
||||
@asynccontextmanager
|
||||
async def lifespan(app: FastAPI):
|
||||
# Startup
|
||||
print("Starting OAK-D service...")
|
||||
"""Startup and shutdown handling."""
|
||||
global running, detection_thread
|
||||
|
||||
print("🦊 Starting OAK-D Vision Service...")
|
||||
|
||||
if init_oak():
|
||||
print("OAK-D initialized successfully!")
|
||||
# Start detection thread
|
||||
running = True
|
||||
detection_thread = threading.Thread(target=detection_loop, daemon=True)
|
||||
detection_thread.start()
|
||||
print("✅ OAK-D service ready!")
|
||||
else:
|
||||
print("Warning: OAK-D not available")
|
||||
print("⚠️ OAK-D not available - running in degraded mode")
|
||||
|
||||
yield
|
||||
|
||||
# Shutdown
|
||||
print("Shutting down OAK-D service...")
|
||||
print("👋 Shutting down OAK-D service...")
|
||||
cleanup_oak()
|
||||
|
||||
|
||||
app = FastAPI(
|
||||
title="OAK-D Vision Service",
|
||||
description="Vixy's eyes! 🦊👀",
|
||||
version="0.1.0",
|
||||
description="Vixy's eyes with face detection! 🦊👀",
|
||||
version="0.2.0",
|
||||
lifespan=lifespan
|
||||
)
|
||||
|
||||
|
||||
# ============== Endpoints ==============
|
||||
|
||||
@app.get("/health")
|
||||
async def health():
|
||||
"""Health check endpoint."""
|
||||
return {
|
||||
"status": "healthy",
|
||||
"service": "oak-service",
|
||||
"version": "0.2.0",
|
||||
"oak_connected": oak_device is not None,
|
||||
"face_detection": detection_queue is not None,
|
||||
"timestamp": time.time()
|
||||
}
|
||||
|
||||
|
||||
@app.get("/presence")
|
||||
async def presence():
|
||||
"""
|
||||
Get current presence state.
|
||||
|
||||
Returns whether someone (Foxy!) is present based on face detection.
|
||||
"""
|
||||
return {
|
||||
"present": presence_state["present"],
|
||||
"face_count": presence_state["face_count"],
|
||||
"last_seen": presence_state["last_seen"],
|
||||
"seconds_since_seen": (
|
||||
time.time() - presence_state["last_seen"]
|
||||
if presence_state["last_seen"] else None
|
||||
),
|
||||
"confidence": presence_state["confidence"],
|
||||
"timestamp": time.time()
|
||||
}
|
||||
|
||||
|
||||
@app.get("/face")
|
||||
async def face():
|
||||
"""
|
||||
Get detailed face detection results.
|
||||
|
||||
Returns bounding boxes and confidence for all detected faces.
|
||||
"""
|
||||
return {
|
||||
"face_count": presence_state["face_count"],
|
||||
"detections": presence_state["detections"],
|
||||
"last_detection": presence_state["last_detection"],
|
||||
"timestamp": time.time()
|
||||
}
|
||||
|
||||
|
||||
@app.get("/snapshot")
|
||||
async def snapshot():
|
||||
"""Capture a single frame from OAK-D RGB camera."""
|
||||
global queue
|
||||
global rgb_queue
|
||||
|
||||
if queue is None:
|
||||
if rgb_queue is None:
|
||||
raise HTTPException(status_code=503, detail="OAK-D not initialized")
|
||||
|
||||
try:
|
||||
frame = queue.get()
|
||||
frame = rgb_queue.tryGet()
|
||||
if frame is None:
|
||||
raise HTTPException(status_code=503, detail="No frame available")
|
||||
|
||||
img = frame.getCvFrame()
|
||||
|
||||
# Encode as JPEG
|
||||
@@ -95,45 +274,65 @@ async def snapshot():
|
||||
content=jpeg.tobytes(),
|
||||
media_type="image/jpeg"
|
||||
)
|
||||
except HTTPException:
|
||||
raise
|
||||
except Exception as e:
|
||||
raise HTTPException(status_code=500, detail=f"Capture failed: {e}")
|
||||
|
||||
|
||||
@app.get("/snapshot/info")
|
||||
async def snapshot_info():
|
||||
"""Capture frame and return metadata without image."""
|
||||
global queue
|
||||
"""Get frame metadata without capturing full image."""
|
||||
global rgb_queue
|
||||
|
||||
if queue is None:
|
||||
if rgb_queue is None:
|
||||
raise HTTPException(status_code=503, detail="OAK-D not initialized")
|
||||
|
||||
try:
|
||||
frame = queue.get()
|
||||
frame = rgb_queue.tryGet()
|
||||
if frame is None:
|
||||
return {"available": False, "timestamp": time.time()}
|
||||
|
||||
img = frame.getCvFrame()
|
||||
|
||||
return {
|
||||
"available": True,
|
||||
"width": img.shape[1],
|
||||
"height": img.shape[0],
|
||||
"channels": img.shape[2],
|
||||
"channels": img.shape[2] if len(img.shape) > 2 else 1,
|
||||
"timestamp": time.time()
|
||||
}
|
||||
except Exception as e:
|
||||
raise HTTPException(status_code=500, detail=f"Capture failed: {e}")
|
||||
raise HTTPException(status_code=500, detail=f"Info failed: {e}")
|
||||
|
||||
|
||||
@app.get("/status")
|
||||
async def status():
|
||||
"""Get OAK-D device status."""
|
||||
"""Get comprehensive OAK-D and presence status."""
|
||||
if oak_device is None:
|
||||
return {"connected": False, "message": "OAK-D not connected"}
|
||||
return {
|
||||
"connected": False,
|
||||
"message": "OAK-D not connected",
|
||||
"presence": presence_state
|
||||
}
|
||||
|
||||
try:
|
||||
return {
|
||||
"connected": True,
|
||||
"device_id": oak_device.getMxId(),
|
||||
"usb_speed": str(oak_device.getUsbSpeed()),
|
||||
"face_detection_enabled": True,
|
||||
"detection_model": FACE_DETECTION_MODEL,
|
||||
"presence": presence_state,
|
||||
"timestamp": time.time()
|
||||
}
|
||||
except Exception as e:
|
||||
return {"connected": False, "error": str(e)}
|
||||
return {
|
||||
"connected": False,
|
||||
"error": str(e),
|
||||
"presence": presence_state
|
||||
}
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
import uvicorn
|
||||
|
||||
6
requirements.txt
Normal file
6
requirements.txt
Normal file
@@ -0,0 +1,6 @@
|
||||
depthai
|
||||
blobconverter
|
||||
opencv-python
|
||||
fastapi
|
||||
uvicorn
|
||||
numpy
|
||||
Reference in New Issue
Block a user