/Real-time_embedded_tracker

A real-time tracker (around 16 fps) for aerial videos for NVIDIA Jetson TX-1

Primary LanguagePython

Real-time_embedded_tracker

A real-time tracker (around 16 fps) for aerial videos for NVIDIA Jetson TX-1

Built over the existing tracker

Hierarchical Convolutional Features for Visual Tracking (ICCV 2015) https://github.com/jbhuang0604/CF2

Usage: python run_tracker.py video_name

Download example data folder at https://drive.google.com/open?id=0B80pwDELxpt5cThjNHk5c1pUVW8 Create data folder in main directory and paste contents.

Directions for creating input:

  1. Convert video into frames.
  2. Create a folder in data directory with name as video_name.
  3. Create a folder in video_name directory called img
  4. Move all images into img directory ( .jpg or .png)
  5. Create a grountruth_rect.txt file in video_name directory.
  6. Enter a single line in groundtruth_rect.txt as follows: x y w h where x = leftmost corner x position y = leftmost corner y position w = width of object h = height of object

Additional files: Keep in main folder

Pretrained CNN model: VGG_CNN_S_deploy.prototxt https://gist.github.com/ksimonyan/fd8800eeb36e276cd6f9#file-vgg_cnn_s_deploy-prototxt

Pretrained CNN prototxt : VGG_CNN_S.caffemodel http://www.robots.ox.ac.uk/~vgg/software/deep_eval/releases/bvlc/VGG_CNN_S.caffemodel

Output:

The tracking output will be shown in a window and frames per second will be printed at the end of processing the entire video.