# General Parameters

{% hint style="info" %}
Use `bash celantur --help`  to print all the parameters and exit.
{% endhint %}

## Anonymisation

<table><thead><tr><th width="324">Parameter</th><th>Description</th></tr></thead><tbody><tr><td><p><code>-a OBJECT</code> </p><p><code>--anonymise OBJECT</code> </p></td><td><p><strong>REQUIRED.</strong></p><p>Select what you want to anonymise. </p><p>Options: <code>face</code>, <code>license-plate</code>, <code>person</code>, <code>vehicle</code></p></td></tr><tr><td><code>--method METHOD</code></td><td><p>Choose anonymisation method.</p><p>Options:</p><ul><li><code>blur</code> : (default) non-reversible blur.</li><li><code>blacken</code> : remove colour value from detection.</li><li><code>pixelate</code> : low-res rendering of detection.</li><li><code>detect</code> : no anonymation.</li></ul></td></tr><tr><td><code>--bbox</code></td><td>Anonymise whole bounding boxes instead of segmentation.</td></tr><tr><td><code>--debug</code></td><td>Visualise bounding boxes, anonymisation and <a href="#format-tiling">tiling</a>.</td></tr><tr><td><p><code>--score</code></p><p><code>--confidence</code></p></td><td>Print the confidence scores for the detected objects on images (only if <code>--debug</code> is enabled).</td></tr><tr><td><code>--save-mask TYPE</code> </td><td><p>Save binary mask or instance mask or both (all).</p><p>See <a data-mention href="/pages/omGhPCeIYcpHVDD8Y0zg">/pages/omGhPCeIYcpHVDD8Y0zg</a>.<br>Options: <code>binary</code>, <code>instance</code>, <code>all</code></p></td></tr><tr><td><code>--mask-scale [0..100]</code></td><td>Downscale binary or segmentation mask (only if <code>--save-mask</code>).</td></tr><tr><td><code>--overwrite</code></td><td>Overwrite existing files in output folder. If not specified, files in input folder will be skipped if file with same name exists in output folder.</td></tr><tr><td><code>--quality [0..100]</code></td><td>Quality of output JPEG image.</td></tr><tr><td><p><code>-f FORMAT</code></p><p><code>--format FORMAT</code></p></td><td>See <a data-mention href="#format-tiling">#format-tiling</a></td></tr><tr><td><p><code>--file-type {image,video}</code></p><p><code>-t {image,video}</code></p></td><td>Select file type for processing. Default: only images selected. Since v26.01.1.</td></tr></tbody></table>

### Blurring gradients

For details see [Customize Blurring](/container/usage/customize-blurring.md)

<table><thead><tr><th width="397">Parameter</th><th>Description</th></tr></thead><tbody><tr><td><code>--face-anonymization-gradient-start</code></td><td>Margin of gradual face blurring</td></tr><tr><td><code>--face-anonymization-gradient-stop</code></td><td>Padding of gradual face blurring</td></tr><tr><td><code>--license-plate-anonymization-gradient-start</code></td><td>Margin of gradual license plate blurring</td></tr><tr><td><code>--license-plate-anonymization-gradient-stop</code></td><td>Padding of gradual license plate blurring</td></tr><tr><td><code>--kernel-size-face</code></td><td>Kernel size for face blur. </td></tr><tr><td><code>--kernel-size-person</code></td><td>Kernel size for person blur. </td></tr><tr><td><code>--kernel-size-license-plate</code></td><td>Kernel size for license plate blur.</td></tr><tr><td><code>--kernel-size-vehicle</code></td><td>Kernel size for vehicle blur.</td></tr></tbody></table>

### Ignore areas

You can define one or multiple rectangular areas within an image that should not be anonymised with [input JSON](/container/usage/batch-and-stream-mode.md#ignore-areas-via-input-json) in Batch and stream mode and with the `ignores` attribute in [REST API mode](/container/usage/rest-api-v1-mode.md#upload-file).

## Inference

<table><thead><tr><th width="336">Parameter</th><th>Description</th></tr></thead><tbody><tr><td><p><code>-e INFERENCE_ENGINE</code></p><p><code>--inference-engine INFERENCE_ENGINE</code></p><p><code>--model INFERENCE_ENGINE</code></p></td><td>Select inference model. Current default is <code>segmentatio-omnium</code> </td></tr><tr><td><code>--face-threshold [0..1]</code></td><td>Threshold for face detection. Defaul: <code>0.2</code>.</td></tr><tr><td><code>--license-plate-threshold [0..1]</code></td><td>Threshold for license plate detection. Default: <code>0.2</code>.</td></tr><tr><td><code>--vehicle-threshold [0..1]</code></td><td>Threshold for vehicle detection. Default: <code>0.4</code>.</td></tr><tr><td><code>--person-threshold [0..1]</code></td><td>Threshold for person detection. Default: <code>0.4</code>.</td></tr><tr><td><code>--cpu-mode</code></td><td>Disable GPU support, <a data-mention href="/pages/Y5Qy2XKl1Dzb69Aei1vH">/pages/Y5Qy2XKl1Dzb69Aei1vH</a>.</td></tr></tbody></table>

## Video processing

<table><thead><tr><th width="265">Parameter</th><th>Description</th></tr></thead><tbody><tr><td><code>--video</code></td><td>Anonymise videos, skip images. Alias for <code>-t video</code> and no <code>-t image</code> </td></tr><tr><td><code>--no-object-tracking</code></td><td>Disables <a data-mention href="/pages/lJ3DUakQZ80cWR6DtRIr">/pages/lJ3DUakQZ80cWR6DtRIr</a> which reduces flickering of anonymized objects in videos.</td></tr><tr><td><code>--keep-bit-rate</code> </td><td>Using this flag will encode the anonymised video with a bitrate close to the original video and ensure a nearly identically file size of the processed video, at the cost of 10% to 20% higher processing time.</td></tr><tr><td><code>--no-ffmpeg-merge</code></td><td><p>Usually after processing the frames, in the post-processing process,</p><ul><li>audio and data streams of the input video is merged into the output video, and</li><li>the bitrate of the output video is set to be close to the input video if <code>--keep-bit-rate</code> .</li></ul><p>With <code>--no-ffmpeg-merge</code>, you disable this step.<br>Since v26.01.1.</p></td></tr></tbody></table>

## Format / Tiling

By specifying image format settings with `--format FORMAT` , `-f FORMAT`, Celantur Container can achieve improved results and apply anonymization only to a specific region of an image.

{% hint style="info" %}
**Don't use spaces in the format string.** \
Otherwise Bash has problems interpreting it as one argument.
{% endhint %}

{% tabs %}
{% tab title="Pre-defined format" %}

You can use the option `--format` to choose the resolution of the input images, eg. `--format pano:8000` for an image resolution of 8000x4000.\
Use `--format whole` (default) for all formats not listed below, or if the input images have different resolutions.

<table><thead><tr><th width="201">Parameter</th><th>Resolution</th></tr></thead><tbody><tr><td><code>pano:4096</code></td><td>4096x2048</td></tr><tr><td><code>pano:5400</code></td><td>5400x2700</td></tr><tr><td><code>pano:5640</code></td><td>5640x2816</td></tr><tr><td><code>pano:7060</code></td><td>7060x3530</td></tr><tr><td><code>pano:7680</code></td><td>7680x3840</td></tr><tr><td><code>pano:8000</code></td><td>8000x4000</td></tr><tr><td><code>pano:7680</code></td><td>7680x3840</td></tr><tr><td><code>pano:8000</code></td><td>8000x4000</td></tr><tr><td><code>pano:8192</code></td><td>8192x4096</td></tr><tr><td><code>pano:11000</code></td><td>11000x5500</td></tr><tr><td><code>whole</code></td><td><strong>default</strong>, valid for all resolutions</td></tr></tbody></table>

{% hint style="info" %}
Ensure that the image resolution matches exactly the predefined resolution in the parameter to avoid processing errors.
{% endhint %}
{% endtab %}

{% tab title="Custom tiling" %}

### Custom tiling for improved results

In certain cases, its beneficial to process high-resolution imagery in tiles, instead of the whole image. Small, distant objects are more likely to be detected.

```bash
--format '{"number":[nx,ny],"overlap":[ox,oy]}'
```

<table><thead><tr><th width="137">Parameter</th><th>Description</th></tr></thead><tbody><tr><td>nx</td><td>Number of horizontal tiles</td></tr><tr><td>ny</td><td>Number of vertical tiles</td></tr><tr><td>ox</td><td>Horizontal overlap between tiles in pixels</td></tr><tr><td>oy</td><td>Vertical overlap between tiles in pixels</td></tr></tbody></table>

### Limit anonymization to a section of the image

Apply anonymization only to a specified rectangular section of the image. This can result in faster processing time.

<figure><img src="/files/9XFJTBNSyxNBz7UwDU5k" alt=""><figcaption></figcaption></figure>

<pre><code><strong>--format '{"number":[3,2],"overlap":[ox,oy],"section":[x1,y1,x2,y2]}'
</strong></code></pre>

{% hint style="info" %}
`number` and `overlap` attribute are required when specifying the section, as seen in the example above.
{% endhint %}

<table><thead><tr><th width="136">Parameter</th><th>Description</th></tr></thead><tbody><tr><td>x1, y1</td><td>Top left coordinate of the section</td></tr><tr><td>x2, y2</td><td>Bottom right coordinate of the section</td></tr></tbody></table>
{% endtab %}
{% endtabs %}

## Configuration (under the hood)

<table><thead><tr><th width="268">Parameter</th><th>Description</th></tr></thead><tbody><tr><td><p><code>-l FILE</code></p><p><code>--license FILE</code>  </p></td><td>Path of the license file.</td></tr><tr><td><p><code>-i PATH</code></p><p><code>--input PATH</code></p><p><code>--input-dir PATH</code></p></td><td>Input file or directory.</td></tr><tr><td><p><code>-o PATH</code> </p><p><code>--output PATH</code> </p><p><code>--output-dir PATH</code></p></td><td>Output directory.</td></tr><tr><td><code>--log-level LEVEL</code></td><td><p>Logging level. </p><p>Options: <code>DEBUG</code>, <code>DETAIL</code>, <code>INFO</code>, <code>WARNING</code>, <code>ERROR</code>, <code>CRITICAL</code></p></td></tr><tr><td><code>--log-dir FOLDER</code></td><td>Location of the log files.</td></tr><tr><td><code>--pid-file FILE</code></td><td>Path  of PID file.</td></tr><tr><td><code>--monitor FILE.json</code></td><td>[EXPERIMENTAL] Gather hardware utilisation in <code>FILE.json</code></td></tr></tbody></table>


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://doc.celantur.com/container/usage/general-parameters.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.