simaaisrc

The simaaisrc GStreamer plugin is a source element that reads data from a file or multiple files specified via properties and pushes it into a GStreamer pipeline. It supports reading data from a single file or multiple files using a format specifier in the location property. Alternatively it can read from multiple files specified through the segments property. The plugin is designed to work with SiMa.ai memory allocators and buffer pools, enabling efficient handling of large data buffers.

Properties

Property	Type	Default	Description
location	string	NULL	Path to the input file (or a format string when multiple files are expected).
index	int	0	Starting file index when using format specifier in location property.
node-name	string	NULL	Output node name used in custom metadata.
mem-target	int	0	Specifies the memory target (0: generic, 1: EV74, 2: DMS0).
delay	int64	0	Milliseconds to sleep between buffer creations.
stream-id	string	“unknown-stream-id”	Stream ID or Camera ID, used in custom metadata.
pool-size	uint	2	Number of buffers allocated in the output buffer pool.
loop	boolean	FALSE	If TRUE, loops back to the initial index after reaching the last file.
segments	string	NULL	Comma separated key=value pairs where key is segment name and value is file path.

Usage

The simaaisrc plugin is integrated into a GStreamer pipeline as a source element. It can be configured using properties.

Example using the ``location`` property to read from a single file:

gst-launch-1.0 simaaisrc location=/tmp/input.bin node-name=mynode ! fakesink dump=true

Example using the ``location`` property and index to read multiple files:

gst-launch-1.0 simaaisrc location=/tmp/input%d.bin index=1 node-name=mynode loop=true ! fakesink dump=true

Example using the ``segments`` property to read data from multiple files:

gst-launch-1.0 simaaisrc segments="hm_tensor=/data/hm_tensor.bin,paf_tensor=/data/paf_tensor.bin" node-name=mynode mem-target=1 ! fakesink dump=true

Installation

To install the simaaisrc plugin, copy the plugin files (from /usr/local/simaai/plugin_zoo/gst-simaai-plugins-base/gst/) into your GStreamer plugins directory (usually plugins/).

Integrating into a GStreamer Pipeline

The simaaisrc element reads data from the specified file(s) and pushes it as GstBuffer objects downstream in the pipeline. Each buffer contains custom metadata including:

• buffer-id: Physical memory address of the buffer.

• buffer-name: Name of the node (specified via the node-name property).

• buffer-offset: Offset within the buffer.

• frame-id: Sequential frame ID.

• stream-id: Stream ID (specified via the stream-id property).

• timestamp: Timestamp of when the buffer was created.

The plugin uses SiMa.ai memory allocation, therefore make sure your build system has appropriate dependencies and linker paths configured. The output buffer’s memory will reflect the memory target specified through the mem-target property.