Skip to content

Latest commit

 

History

History
131 lines (76 loc) · 3.72 KB

README.md

File metadata and controls

131 lines (76 loc) · 3.72 KB

#Installation Instructions: Download scripts in an existing ROS package

Download needed libraries:

sudo apt-get install python-scipy python-sklearn

Tested on R.O.S. Hydro, Ubuntu 12.04

Python v. 2.7.3

Scipy v. 0.9.0

Sklearn v. 0.10

laser scanner f.o.v : +- 45 deg. , intensity publishing enabled(not used yet)

Human-Pattern-Recognition

Real time recongition of humans through laser scans

#Sample Run 1)Record bag file with laser scans (.bag files are provided in video folder)

2)Open a terminal and run 
	roscore

3)On another terminal 
	cd 
  to the directory of the scripts

4)Convert a .bag file to .mat format by running 
	python bag2mat.py video/video2.bag video/video2.mat scan 79

4.1)Convert as many as you want by changing the number of the video .bag file

5)Manually annotate the data by running 
	python annotate.py 40 10 video/video2.mat

5.1)Annotate as many as you want by changing the number of the video .mat file

6)Create a classifier, and P.C.A. object by running 
	python merge_train.py video/

7)Test online with the previously created classifier by running 
	python hpr.py video/Gaussian_NB_classifier_merged.p video/PCA_object.p scan 40 10 0

8)In another terminal 
	cd 
  to the directory of the scripts.

9)Run 
	rosbag play video/video10.bag 
  so that the script in the previous step is triggered.

#Python files explained #----------------------

#I)bag2mat.py:

Convert a .bag file to .mat format.

Command line use:

python bag2mat.py <.bag_file_path> <.mat_file_path> <laser_scan_rostopic> <scan_duration>

#II)annotate.py :

Annotate (label) each cluster as human or not human, in order to train the classifier later. annotate.py generates many .p files, some of which are classifiers trained on each and only file annotated. This means that every time the annotation process ends, a new classifier is created trained only on the .mat that was just annotated.

Command line use:

python annotate.py <time_window> <number_of_frames_to_create_walls> <.mat_file_path>

#III)merge_train.py:

Merge_train uses all the annotations from a folder, to create a classifier based on all of those annotations.

Command line use:

python merge_train <folder of annotated .mat files>

#IV)hpr.py:

Runs the human pattern recognition (Naive Bayes) classifier.

Command line use:

python hpr.py <classifier_object_path> <pca_objec_path> <laserscan_topic> <timewindow_in_frames> <maximum_scan_range> <0_or_1_for_metrics>

#V)generate_metrics.py

Runs the mat file that the classifier was tested on, and lets the user annotate the same clusters as the classifier in order to generate some basic metrics (Precision, Recall, Accuracy).

Command line use :

python generate_metrics.py </path/to/classification_results.mat>

#VI)hpr_mat.py : Runs the specified classifier on a .mat file instead of a topic.

Command line use :

python hpr_mat.py <.mat_file_path> <classifier_path> <pca_object_path> <timewindow> <frames for walls> <0_or_1_for_metrics>

#VII)Library files gridfit.py, myhog.py and mytools.py are libraries used to run the code. They were pre-written by others or ported to python.

#VIII)online_calibration_node.py, optimal_test.py These two python files are not used anymore. They are still here just for legacy reasons and in case something new comes to our mind. You can freely ignore them.

#IX)ideal_data.mat This .mat file contains the basic information regarding the laser scans. It is not used by any of our currently used scripts, but it's still here for the sake of completeness and future reference.

RECOMMENDATIONS:

1) Run the Sample Run first. It will help you understand what each .py does.
2) Use same timewindow, and wall set time for each training set, and use the same values when
evaluating.