Browser Surveillance Station

Use it when observation is needed now, but installing an Agent is not possible

Browser Surveillance Station runs as an experimental use case in the Banalytics Console. It asks the browser for access to a local camera and, only when selected, a microphone; it performs simple motion analysis in the open browser session and creates short clips when motion is detected or when an authorised Telegram user requests one.

It is useful for a rapid, local proof of concept-not for unattended, long-term, or safety-critical monitoring. There is no installed runtime, durable recording service, managed camera reconnect, guaranteed retention, or background execution once the browser session is suspended or closed.

The security and data boundaries are described below. The person who opens the component controls camera permission, starts and stops capture, decides whether messenger delivery is enabled, supplies the API token, approves or removes chats, and can revoke that delivery path. Banalytics does not initiate a clip transfer or select a recipient. An approved chat can request a clip only during the active session and only because the user explicitly granted that chat access.

Do not use this component as the only control for life safety, security dispatch, access control, employee monitoring, regulatory evidence retention, or any workflow that requires continuous coverage. Those needs call for an assessed production design with an installed Edge Agent or another managed video platform.

Four focused panels keep the observation workflow visible

The screen is arranged around the live scene. The camera preview shows the browser-selected source; the recordings area keeps a short, in-session list of generated clips; the right column contains four collapsible panels so an operator can keep only the controls needed for the current task open.

Each right-column panel can be collapsed or expanded. This is especially useful on smaller screens: start with the camera and the one setting you need, then open Telegram delivery only when a notification path is required.

Choose the source and set a bounded recording envelope

1. Grant browser permission. Select the intended camera and, when audio is needed, microphone in the browser permission prompt. Camera labels are often visible only after permission is granted.
2. Select video and audio sources. Use a front or rear phone camera, a laptop webcam, a USB camera, or another browser-visible input. Verify the preview after every source change.
3. Choose a practical resolution. Start with 640×480. Increase resolution only when the evidence needs more detail and the device can sustain it without dropped frames, overheating, or excessive battery use.
4. Set "Store video of greater than". This filters automatic motion clips shorter than the chosen number of seconds. It helps suppress very brief triggers. Explicit Telegram video-shot requests are retained even when short.
5. Set "Split long videos to the duration". This is the maximum duration of a single recording segment. Shorter segments are easier to review and send, and they place less sustained pressure on a mobile device.

Configuration is session-oriented. Treat it as a field setup: write down the purpose, camera position, operator, start time, intended recipients, and any limits agreed for the proof of concept. If the observation becomes recurring, migrate those settings into a managed deployment rather than relying on a browser tab.

Configure the motion detector

The preview image shows the parts of the scene where the background has changed. A person passing through the frame, an opening door, moving equipment, a shadow, or camera noise can all produce a changed region.

The histogram below the image helps make that visible. It shows how often the detector sees small, medium, or large changed regions: a cluster of small values is usually harmless background noise, while a larger region is more likely to be the movement the operator cares about. The slider sets the minimum region size that starts a reaction and recording.

• Start by ignoring regular small noise. Move the slider above the small values that appear constantly, then check that the intended movement still creates a larger region.
• Test the actual scene. Light changes, reflections, rain, vibration, autofocus, and activity at the edge of the frame can change the result.
• Review a few clips. Adjust the slider until the recordings contain useful movement without filling the list with noise.

A practical proof of concept should define the observation in one sentence: for example, "notify the duty supervisor when a vehicle-sized movement reaches the temporary loading bay." If that sentence cannot name a reviewer and response, the component is collecting video without a clear operational purpose.

Connect only a bot and chats that are authorised for the task

Telegram delivery is optional. When enabled, the browser connects to the Telegram Bot API with the token supplied by the operator. A new chat is not allowed automatically: the operator must confirm the access request in the browser. The Telegram Delivery panel always shows the allowed-user list, including when the bot is not currently active, so access can be reviewed before the next session.

1. Create or select the Telegram bot. Obtain its API token from Telegram and enter it in the component. Treat the token as a secret with the same care as a password for that bot.
2. Decide whether to remember the token. Enabling Remember token displays a confirmation explaining that the token is stored in the browser’s local storage on that device, for the console.banalytics.live origin, under browser-surveillance-station.telegram_bot_token. It is not a suitable option for a shared or untrusted device.
3. Start the bot and approve only intended chats. When a person first messages the bot, the local operator confirms or rejects the chat. Approved chat identifiers are shown in the panel and can be removed with the disconnect control.
4. Use the Telegram buttons deliberately. After a chat is approved, the bot sends that chat interactive /videoshot 1, /videoshot 5, /videoshot 10, /videoshot 30, and /videoshot 60 buttons. Selecting one creates a clip of the chosen duration and delivers it to the allowed chats while the bot is active.
5. Use the matching bot command when appropriate. An allowed Telegram user may send /videoshot or /videoshot <seconds>. The browser creates the requested short clip; it does not give the user general control of the device.

Removing the check from Remember token immediately deletes the saved token and saved allowed/denied chat configuration from that browser’s local storage. If the bot is already running, it keeps the token only in its active browser memory until the user stops it; it is not saved for the next session.

A browser tab is an interactive foreground tool, not an always-on recorder

Browser media APIs are intentionally constrained to protect battery, privacy, and device resources. Exact behaviour depends on the browser, operating system, permissions, hardware, and enterprise device-management policies. Validate the target device before relying on the component in a real situation.

These limitations are the reason an installed Banalytics Edge Agent is the appropriate next step for continuous observation. An Agent can be deployed on a controlled host with defined power, storage, retention, monitoring, restart behaviour, and access policy; the browser component cannot provide those operational guarantees.

Keep the data path explicit before granting camera access

When the user opens this component, the Banalytics Console downloads unobfuscated, inspectable JavaScript and OpenCV support into that user’s browser. The code runs inside the browser process on the user’s device. Banalytics personnel do not start the camera, operate the browser, or receive the live stream through this component.

The user’s browser asks the user for camera permission and, only when selected, microphone permission. After permission is granted, the user’s device acquires the selected stream and performs motion processing in that browser tab. The component does not upload the live camera stream to a Banalytics video backend as part of its normal operation.

Controller of the observation data: the organisation or individual using the component controls the resulting video and personal data. That party chooses the camera, purpose, location, recipients, retention, API token, and messenger bot; it must establish a lawful purpose, provide any required notices, limit access, set retention, respond to data-subject requests, and decide whether the selected messenger is an approved recipient.

Banalytics’ role: Banalytics delivers the inspectable code to the user’s browser. Banalytics does not operate the user’s camera, select recipients, start clip delivery, or receive the live stream through this component. Banalytics may separately process Console-account and service telemetry under the applicable service relationship; that is distinct from the browser component’s local camera stream.

Messenger’s role: once the user’s browser sends a clip, the selected messenger is an independent service in the delivery path. Use only a bot, chats, and recipient devices approved by the deployment owner. Do not send sensitive video merely because the technical connection works.