By Jeremy Koski · September 17, 2026
Security cameras are one of the most common reasons people set up Home Assistant. But buying cameras that actually work well with Home Assistant requires knowing which protocols to look for, how storage works, and what to avoid. This guide covers the practical questions you need to answer before spending money on cameras for a Home Assistant setup.
Home Assistant supports several camera types, but not all of them work equally well. The best cameras for Home Assistant are ones that provide a local RTSP or MJPEG stream. RTSP (Real-Time Streaming Protocol) is the standard for IP cameras and gives Home Assistant a direct video feed without relying on a cloud service. MJPEG is an older alternative that works but uses more bandwidth.
Cameras that only work through a manufacturer's cloud API (like Ring, Nest, or Wyze in default mode) are harder to integrate. Some have Home Assistant integrations through unofficial channels, but these can break when the manufacturer changes their API. If your goal is a reliable, local-only system, choose cameras that advertise RTSP support.
This is one of the first decisions to make.
WiFi cameras are easier to install because you do not need to run Ethernet cables. You plug them into a power outlet, connect to WiFi, and configure the stream. The downside is that WiFi is less reliable than wired. Video streams are bandwidth-heavy, and multiple WiFi cameras on a single access point can cause congestion and dropped frames. WiFi cameras also require a nearby power outlet, limiting placement options.
PoE cameras receive both power and data through a single Ethernet cable. You need a PoE switch or PoE injectors to power them, but once the cable is run, you have a reliable wired connection with no WiFi dependency. PoE cameras are the standard for anyone serious about security camera setups. They are more reliable, easier to manage on the network, and can be placed anywhere you can run a cable. If you are installing more than two cameras, PoE is the right choice.
Cloud storage means your camera footage is sent to a manufacturer's server. You pay a monthly fee, and you access recordings through their app. The problem is that if the company changes their pricing, shuts down, or changes their API, your recordings may become inaccessible or your Home Assistant integration may break.
Local storage means recordings stay on your network. You can use an SD card in the camera, a local NVR (Network Video Recorder), or a self-hosted NVR like Frigate running on your Home Assistant server. Local storage gives you full control, no monthly fees, and no dependency on a third-party service. For a Home Assistant setup, local storage is the right approach.
Resolution determines how much detail you can see. Common options are 1080p (2MP), 4MP (1440p), and 8MP (4K). Higher resolution means more detail but also more storage and bandwidth. For most home security use, 4MP is a good balance. It provides enough detail to identify faces at typical viewing distances without consuming storage as fast as 4K.
Night vision matters because most security incidents happen at night. There are two approaches: infrared (IR) LEDs and color night vision with a sensor and supplemental light. IR night vision produces black-and-white images in the dark. Color night vision requires some ambient light or a dedicated light source but gives you color information that can be useful for identifying vehicles and clothing. For a front door or driveway, color night vision with a motion-activated light is worth the extra cost.
ONVIF is an open standard for IP cameras. When a camera supports ONVIF Profile S, it means it provides a standard RTSP stream that Home Assistant can connect to without needing a manufacturer-specific integration. ONVIF Profile T adds advanced features like two-way audio and event streaming.
Not all cameras that claim ONVIF support implement it correctly. Before buying, check the Home Assistant community forums for reports on specific models. A camera with ONVIF Profile S and positive Home Assistant user reports is a safe bet. Cameras without ONVIF may still work through manufacturer-specific integrations, but you depend on that integration being maintained.
Frigate is an open-source NVR designed specifically for Home Assistant. It runs as an add-on or a Docker container alongside Home Assistant and provides motion detection, object detection, and recording. Frigate uses a machine learning model to detect people, cars, and other objects, which means you get fewer false alerts than a simple motion sensor. It supports Google Coral or OpenVINO accelerators for fast object detection without loading your CPU.
With Frigate, you set up cameras, define zones for detection, and configure recording to a local drive. Detected objects can trigger automations, send notifications, or control other devices through Home Assistant. If you want a security camera system that runs entirely on your own hardware, Frigate is the recommended approach. For setup guidance, see our post on running Home Assistant on a Raspberry Pi in 2026, which covers hardware requirements that apply to Frigate as well.
Security cameras record everything in their field of view. Cloud-based cameras send footage to third-party servers, which means the manufacturer and potentially law enforcement can access your recordings. Local storage keeps footage on your network, but you are still responsible for securing access to it. Use strong passwords on your cameras, disable cloud features when possible, keep your Home Assistant instance behind a VPN (see our guide on smart home security for details), and avoid placing indoor cameras in private spaces.
Camera prices vary widely. Here is what to expect:
Where you put cameras matters as much as which cameras you buy:
Avoid pointing cameras at neighbors' windows or public sidewalks. Motion detection will trigger constantly on passing traffic.
For a Home Assistant security camera system, start with two or three PoE cameras from a mid-range brand like Reolink or Amcrest that support ONVIF Profile S. Run Frigate for recording and object detection, and store footage locally. Add cameras as needed. Avoid cloud-dependent cameras unless you have a specific reason to use one. This gives you a reliable, private system that does not depend on a manufacturer staying in business or keeping their API stable.
If you want help selecting and installing security cameras that work with Home Assistant, see our smart home services page or book an on-site visit and we will handle the full setup.
For more practical technology guides, check out our post on running Home Assistant on a Raspberry Pi in 2026.