tfjs-examples/visualize-convnet/index.html

86 lines
2.7 KiB
HTML

<!--
Copyright 2019 Google LLC. All Rights Reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
==============================================================================
-->
<!doctype html>
<head>
<title>TensorFlow.js: Visualize Convnet Internals</title>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1">
<link rel="stylesheet" href="../shared/tfjs-examples.css" />
<link rel="stylesheet" href="./style.css" />
</head>
<body>
<div class="tfjs-example-container">
<section class='title-area'>
<h1>TensorFlow.js: Visualize Convnet Internals</h1>
</section>
<section>
<p class='section-head'>Description</p>
<p>
This example demonstrates some techniques of visualizing
the internal workings of a convolutional neural network (convnet)
in TensorFlow.js, including:
<ul>
<li>
Finding what convolutional layers' filters are sensitive to after
training: calculating maximally-activating input image for
convolutional filters through gradient ascent in the input space.
</li>
<li>
Getting the internal activation of a convnet by uisng the
functional model API of TensorFlow.js
</li>
<li>
Finding which part of an input image is most relevant to the
classification decision made by a convnet (
<a href="https://keras.io/applications/#vgg16">VGG16</a>
in this case), using the gradient-based class activation map (CAM)
approach.
</li>
</ul>
</p>
</section>
<section>
<p class='section-head image-result-heading'>
Input image and classification result
</p>
<div id="image-result"></div>
<p class='section-head'>Visualization</p>
<div>
<span>What to visualize:</span>
<select id="viz-type">
<option value="activation">Filter activation</option>
<option value="filters">Maximally-activating input images</option>
<option value="cam">Class activation map (CAM)</option>
</select>
</div>
<div id="viz-section"></div>
</section>
<script type="module" src="index.js"></script>
</div>
</body>