page_type: reference description: DataLoader class for Image Searcher Task.

tflite_model_maker.searcher.ImageDataLoader

<table class="tfo-notebook-buttons tfo-api nocontent" align="left"> <td> <a target="_blank" href="https://github.com/tensorflow/examples/blob/master/tensorflow_examples/lite/model_maker/core/data_util/image_searcher_dataloader.py#L33-L155">

View source on GitHub

</a> </td> </table>

DataLoader class for Image Searcher Task.

Inherits From: DataLoader

<pre class="devsite-click-to-copy prettyprint lang-py tfo-signature-link"> <code>tflite_model_maker.searcher.ImageDataLoader( embedder: image_embedder.ImageEmbedder, metadata_type: <a href="../../tflite_model_maker/searcher/MetadataType"><code>tflite_model_maker.searcher.MetadataType</code></a> = <a href="../../tflite_model_maker/searcher/MetadataType#FROM_FILE_NAME"><code>tflite_model_maker.searcher.MetadataType.FROM_FILE_NAME</code></a> ) -> None </code></pre>   <table class="responsive fixed orange"> <colgroup><col width="214px"><col></colgroup> <tr><th colspan="2"><h2 class="add-link">Args</h2></th></tr> <tr> <td> `embedder`<a id="embedder"></a> </td> <td> Embedder to generate embedding from raw input image. </td> </tr><tr> <td> `metadata_type`<a id="metadata_type"></a> </td> <td> Type of MetadataLoader to load metadata for each input data. By default, load the file name as metadata for each input data. </td> </tr> </table>  <table class="responsive fixed orange"> <colgroup><col width="214px"><col></colgroup> <tr><th colspan="2"><h2 class="add-link">Attributes</h2></th></tr> <tr> <td> `dataset`<a id="dataset"></a> </td> <td> Gets the dataset.

Due to performance consideration, we don't return a copy, but the returned self._dataset should never be changed.

</td> </tr><tr> <td> `embedder_path`<a id="embedder_path"></a> </td> <td> Gets the path to the TFLite Embedder model file. </td> </tr><tr> <td> `metadata`<a id="metadata"></a> </td> <td> Gets the metadata. </td> </tr> </table>

Methods

<h3 id="append"><code>append</code></h3>

<a target="_blank" class="external" href="https://github.com/tensorflow/examples/blob/master/tensorflow_examples/lite/model_maker/core/data_util/searcher_dataloader.py#L92-L106">View source</a>

<pre class="devsite-click-to-copy prettyprint lang-py tfo-signature-link"> <code>append( data_loader: 'DataLoader' ) -> None </code></pre>

Appends the dataset.

Don't check if embedders from the two data loader are the same in this function. Users are responsible to keep the embedder identical.

<table class="responsive fixed orange"> <colgroup><col width="214px"><col></colgroup> <tr><th colspan="2">Args</th></tr> <tr> <td> `data_loader` </td> <td> The data loader in which the data will be appended. </td> </tr> </table> <h3 id="create"><code>create</code></h3>

<a target="_blank" class="external" href="https://github.com/tensorflow/examples/blob/master/tensorflow_examples/lite/model_maker/core/data_util/image_searcher_dataloader.py#L60-L92">View source</a>

<pre class="devsite-click-to-copy prettyprint lang-py tfo-signature-link"> <code>@classmethod</code> <code>create( image_embedder_path: str, metadata_type: <a href="../../tflite_model_maker/searcher/MetadataType"><code>tflite_model_maker.searcher.MetadataType</code></a> = <a href="../../tflite_model_maker/searcher/MetadataType#FROM_FILE_NAME"><code>tflite_model_maker.searcher.MetadataType.FROM_FILE_NAME</code></a>, l2_normalize: bool = False ) -> 'DataLoader' </code></pre>

Creates DataLoader for the Image Searcher task.

<table class="responsive fixed orange"> <colgroup><col width="214px"><col></colgroup> <tr><th colspan="2">Args</th></tr> <tr> <td> `image_embedder_path` </td> <td> Path to the ".tflite" image embedder model. </td> </tr><tr> <td> `metadata_type` </td> <td> Type of MetadataLoader to load metadata for each input image based on image path. By default, load the file name as metadata for each input image. </td> </tr><tr> <td> `l2_normalize` </td> <td> Whether to normalize the returned feature vector with L2 norm. Use this option only if the model does not already contain a native L2_NORMALIZATION TF Lite Op. In most cases, this is already the case and L2 norm is thus achieved through TF Lite inference. </td> </tr> </table>  <table class="responsive fixed orange"> <colgroup><col width="214px"><col></colgroup> <tr><th colspan="2">Returns</th></tr> <tr class="alt"> <td colspan="2"> DataLoader object created for the Image Searcher task. </td> </tr> </table> <h3 id="load_from_folder"><code>load_from_folder</code></h3>

<a target="_blank" class="external" href="https://github.com/tensorflow/examples/blob/master/tensorflow_examples/lite/model_maker/core/data_util/image_searcher_dataloader.py#L94-L155">View source</a>

<pre class="devsite-click-to-copy prettyprint lang-py tfo-signature-link"> <code>load_from_folder( path: str, mode: str = 'r' ) -> None </code></pre>

Loads image data from folder.

Users can load images from different folders one by one. For instance,

# Creates data_loader instance.
data_loader = image_searcher_dataloader.DataLoader.create(tflite_path)

# Loads images, first from `image_path1` and secondly from `image_path2`.
data_loader.load_from_folder(image_path1)
data_loader.load_from_folder(image_path2)

<table class="responsive fixed orange"> <colgroup><col width="214px"><col></colgroup> <tr><th colspan="2">Args</th></tr> <tr> <td> `path` </td> <td> image directory to be loaded. </td> </tr><tr> <td> `mode` </td> <td> mode in which the file is opened, Used when metadata_type is FROM_DAT_FILE. Only 'r' and 'rb' are supported. 'r' means opening for reading, 'rb' means opening for reading binary. </td> </tr> </table> <h3 id="__len__"><code>__len__</code></h3>

<a target="_blank" class="external" href="https://github.com/tensorflow/examples/blob/master/tensorflow_examples/lite/model_maker/core/data_util/searcher_dataloader.py#L60-L61">View source</a>