Not clear about whether multi-input networks are supported in the desginer tool of DL toolbox. I have two networks, one CNN which takes images as input, the other numerical input data for a MLP. I can concatenate in desginer and analyze without an issue. How do I configure the two data streams to upload to such a network? (For example, I use imdatastore for the images.) Can the above be done in the designer tool? Currently, I get an error after uploading data and trying to train such a network. Error says it expects only one input. Are there any examples for using such mixed data and training the network?

 採用された回答

Srivardhan Gadila
Srivardhan Gadila 2020 年 10 月 28 日

0 投票

As of current release, Deep Network Designer does not support networks with multiple input layers. You can refer to Multiple-Input and Multiple-Output Networks to implement programatically.
You can also refer to the following MATLAB Answers to get started: How can I train multi-input deep network without DataStore & Multi-Input CNN for image classification and MathWorks Help Center for related information.

その他の回答 (0 件)

カテゴリ

ヘルプ センター および File ExchangeRecognition, Object Detection, and Semantic Segmentation についてさらに検索

質問済み:

2020 年 10 月 21 日

回答済み:

2020 年 10 月 28 日

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by