CODEWITHSUNDEEP
tensorflowdeep-learningconv-neural-networkobject-detection-api
batuman
batuman

Reputation: 7304

What does min_depth in ssdlite_mobilenet_v2_coco.config do?

In Tensorflow's object detection API, ssdlite_mobilenet_v2_coco.config file has min_depth parameter. Now it is set to 16.

Is it minimum network length set to 16?

If I change to lower value, will the network length be shorter? What layers will be removed?

Upvotes: 1

Views: 1030

Answers (1)

Danny Fang
Danny Fang

Reputation: 4071

Multi-resolution feature maps

This parameter stands for the minimum depth of each of the output feature maps produced by the feature extractor. As we know SSD use multi-resolution feature maps to help detect objects of multiple scales. Changing it to a smaller value won't necessarily make the depth shorter as the depth is also controlled by the depth_multiplier as shown in this function. So for each extra feature layer as shown in the figure, the two parameter will control the depth of it. More details are specified in class KerasMultiResolutionFeatureMaps and function multi_resolution_feature_maps

Upvotes: 2

Related Questions