Tensorflow Object Detection

Video and Tensorflow Object Detection API

#Automated Process

This will take the video given, read it into the Tensorflow Object Detection API. It will then create a new video with the objects detected. s You can choose other models to make it find more objects or run faster.

Make sure there is no .DS_Store or other non-image file in the video output (--output) folder.

cp main_single_pass.py models/research/object_detection
 
# Run the application
python main_single_pass.py --video=/path/to/a/video.mp4 --model=/path/to/a/model/folder --output=/path/to/output/new/video

Installation

You can run the install script to install all the dependencies.

./install.sh

This will clone the github Tensorflow Object Detection API. It will install all the dependencies. It will create all the needed folders.

More details can be found here: http://www.sroboto.com/2017/09/pass-video-into-tensorflow-object.html

#Manual Process

Convert video to files

This will convert the video dog_video.mp4 to images in video_output/ folder.

python convert_video_to_images.py

Process the images

This will process the images. The processed images will be in video_output/output

cd models\research\object_detection
python main.py

Convert Processed images back to Video

This will convert the images in video_output/output to a video output.mp4

python convet_images_to_video.py