tfrun

An easy-to-use C++ wrapper over the stable C API of TensorFlow
git clone https://0xff.ir/g/tfrun.git
Log | Files | Refs | README | LICENSE

mobilenet-v2-1.4-224.md (1648B)


      1 # `mobilenet-v2-1.4-224`
      2 
      3 **NOTE** This file is adapted from [here](https://github.com/opencv/open_model_zoo/blob/master/models/public/mobilenet-v2-1.4-224/mobilenet-v2-1.4-224.md).
      4 
      5 ## Use Case and High-Level Description
      6 
      7 `mobilenet-v2-1.4-224` is one of MobileNets - small, low-latency, low-power
      8 models parameterized to meet the resource constraints of a variety of use
      9 cases. They can be built upon for classification, detection, embeddings and
     10 segmentation similar to how other popular large scale models are used.
     11 For details, see the [paper](https://arxiv.org/abs/1704.04861).
     12 
     13 ## Example
     14 
     15 ## Specification
     16 
     17 | Metric                          | Value                                     |
     18 |---------------------------------|-------------------------------------------|
     19 | Type                            | Classification                            |
     20 | GFlops                          | 1.183                                     |
     21 | MParams                         | 6.087                                     |
     22 
     23 ## Performance
     24 
     25 ## Input
     26 
     27 Image, name: `input` , shape: [1x224x224x3], format: [BxHxWxC], where:
     28 
     29     - B - batch size
     30     - H - image height
     31     - W - image width
     32     - C - number of channels
     33 
     34    Expected color order: RGB.
     35    Mean values: [127.5, 127.5, 127.5], scale factor for each channel: 127.5
     36 
     37 ## Output
     38 
     39 Probabilities for all dataset classes (0 class is background). Probabilities
     40 are represented in logits format. Name: `MobilenetV1/Predictions/Reshape_1`.
     41 
     42 ## Legal Information
     43 
     44 The original model is distributed under the
     45 [Apache License, Version 2.0](https://raw.githubusercontent.com/tensorflow/models/master/LICENSE).