The i3d-rgb-tf
is a model for video classification, based on paper "Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset". This model use RGB input stream and trained on Kinetics-400 dataset. Additionally, this model has initialize values from Inception v1 model pre-trained on ImageNet dataset.
Originally redistributed as a checkpoint file, was converted to frozen graph.
- Clone or download original repository:
git clone https://github.com/deepmind/kinetics-i3d.git
- (Optional) Checkout the commit that the conversion was tested on:
git checkout 0667e88
- Install prerequisites, tested with:
tensorflow==1.11 tensorflow-probability==0.4.0 dm-sonnet==1.26
- Copy
<omz_dir>/models/public/i3d-rgb-tf/freeze.py
script to root directory of original repository and run it:python freeze.py
Metric | Value |
---|---|
Type | Action recognition |
GFLOPs | 278.981 |
MParams | 12.69 |
Source framework | TensorFlow* |
Accuracy validations performed on validation part of Kinetics-400 dataset. Subset consists of 400 randomly chosen videos from this dataset.
Metric | Converted Model | Converted Model (subset 400) |
---|---|---|
Top 1 | 65.96% | 64.83% |
Top 5 | 86.01% | 84.58% |
Video clip, name - Placeholder
, shape - 1, 79, 224, 224, 3
, format is B, D, H, W, C
, where:
B
- batch sizeD
- duration of input clipH
- heightW
- widthC
- channel
Channel order is RGB
. Mean value - 127.5, scale value - 127.5.
Video clip, name - Placeholder
, shape - 1, 79, 224, 224, 3
, format is B, D, H, W, C
, where:
B
- batch sizeD
- duration of input clipH
- heightW
- widthC
- channel
Channel order is RGB
.
Action classifier according to Kinetics-400 action classes, name - Softmax
, shape - 1, 400
, format is B, C
, where:
B
- batch sizeC
- predicted probabilities for each class in [0, 1] range
Action classifier according to Kinetics-400 action classes, name - Softmax
, shape - 1, 400
, format is B, C
, where:
B
- batch sizeC
- predicted probabilities for each class in [0, 1] range
You can download models and if necessary convert them into OpenVINO™ IR format using the Model Downloader and other automation tools as shown in the examples below.
An example of using the Model Downloader:
omz_downloader --name <model_name>
An example of using the Model Converter:
omz_converter --name <model_name>
The model can be used in the following demos provided by the Open Model Zoo to show its capabilities:
The original model is distributed under the
Apache License, Version 2.0. A copy of the license is provided in <omz_dir>/models/public/licenses/APACHE-2.0.txt
.