How to Fix Frigate Detection Delays and Missed Objects
- Detect resolution too low to recognize objects
- Detect fps too low
- Object filters too strict (min_area/threshold/min_score)
Problem Description
Frigate is missing objects (people, cars) that walk through the camera view, or detections are delayed by several seconds. People may appear and leave the frame before Frigate identifies them. Low detect FPS, overloaded detection hardware, high detect resolution, overly strict motion filtering, or degraded sub stream quality are the most common causes.
Why This Happens in Real Homes
Frigate missing objects or detecting them late usually comes down to the detect pipeline's settings and load. Detection only runs where motion is found, at the detect stream's resolution and fps, and is then filtered by object thresholds — so a detect resolution too low to recognize a distant person, an overloaded detector adding delay, or overly strict min_area/threshold filters all cause misses.
Make sure the detect resolution is high enough that objects appear large enough to recognize (very low-res sub-streams miss distant objects), and that detect fps and detector capacity are adequate so frames aren't delayed. Loosen object filters (min_area, score thresholds) if they're excluding real detections, and confirm motion detection and zones/masks actually cover the area. Balancing detect resolution, load, and filter strictness catches the objects reliably.
Symptoms
- Missing objects (people/cars)
- Detection delays
- Objects not detected
- Late detections
- Misses obvious objects
- Inconsistent detection
- Detector too slow
- Objects slip through
Recognize these? Here's what usually causes it.
Common Causes
- Detect resolution too low to recognize objects
- Detect fps too low
- Object filters too strict (min_area/threshold/min_score)
- Detector overloaded (delays)
- Motion not triggering detection in the area
- Object too small/far in frame
- Model limitations
- Zones/masks excluding the area
Most fixes happen in the first 3 steps.
Over-masking can hide real events; review masks carefully.
Tools & Requirements
Step-by-Step Solution
Check the detect stream FPS
Frigate processes frames from the detect stream at the configured FPS. If detect FPS is set to 2: Frigate only analyzes 2 frames per second, meaning a person walking quickly through the frame may appear in only 1-2 frames and get missed. Increase detect FPS: cameras: your_camera: detect: fps: 5. The sweet spot for most setups is 5 FPS — high enough to catch fast movement, low enough to keep CPU manageable. Going above 10 FPS rarely improves detection and significantly increases CPU load.
Check CPU or Coral TPU utilization
If the detection hardware (CPU or Coral TPU) is overloaded: frames are dropped before detection runs, causing missed objects. Check the Frigate system dashboard — the inference speed should be under 100ms per frame on Coral TPU, or under 200ms on CPU. If inference takes longer: frames queue up and get dropped. With 6+ cameras at 5 FPS each, that is 30 frames/second of detection work — a single Coral USB can handle this, but CPU-only detection cannot. Add a Coral TPU or reduce the number of cameras with detection enabled.
Reduce the detect stream resolution
Higher resolution detect streams take longer to process. The object detection model runs on a fixed input size (320x320 for SSD MobileNet) — sending 1080p frames means Frigate must downscale every frame before inference, wasting CPU. Set the detect stream to 640x480 or 720p in your camera settings: cameras: your_camera: detect: width: 1280, height: 720. The detection model works equally well at 720p and processes frames significantly faster than 1080p or 4K.
Check motion detection sensitivity
Frigate uses motion detection as a pre-filter — object detection only runs on frames where motion is detected. If the motion threshold is too high: small or slow-moving objects do not trigger motion detection, and the detector never sees them. Lower the motion threshold: motion: threshold: 20 (default is 30, lower = more sensitive). Also check motion contour_area — this sets the minimum moving area in pixels. Reduce it if small objects (pets, people at a distance) are being filtered out before they reach the object detector.
Verify the camera's sub stream quality
The detect stream is typically the camera's sub stream. If the sub stream quality is too low (excessive compression, very low bitrate): the image is too degraded for the detection model to identify objects. Check the camera's sub stream settings: set bitrate to at least 512kbps, use H.264 CBR encoding, and set GOP (keyframe interval) to 1-2 seconds. Very high compression causes blocking artifacts that the AI model cannot interpret — it sees a blocky mess instead of a person.
Quick Solutions
Still having issues? This is usually the deeper cause below.
If the sensor still misses events after repositioning, check whether a scheduled 'home' or 'away' mode is overriding the sensitivity setting silently.
Tune one camera profile first, then clone to similar cameras.
Notification delays over 2 minutes are almost never the device's fault — background app restrictions quietly re-enable themselves after every OS update.
- Detect resolution too low to recognize objects
- Detect fps too low
- Object filters too strict (min_area/threshold/min_score)
- Detector overloaded (delays)
- Motion not triggering detection in the area
Before you go — try one of these (they fix most cases).
Official Manufacturer Manual
Frigate provides official product documentation through their online manual rather than downloadable PDF. Access setup guides, troubleshooting steps, and product specifications for your Frigate Detection Tuning.
Source: docs.frigate.video
Need More Help? Frigate Support
Note: The contact information below connects you directly to Frigate's official customer support team, not Trunetto. They can help with warranty claims, device replacements, and advanced technical issues.