README.md 3.52 KB
Newer Older
dhomm's avatar
dhomm committed
1
2
# Bachelor Thesis - Sensor Localization

dhomm's avatar
dhomm committed
3
4
The finished systmes can be found on the main and on the master branch.
Scripts for the feature evaluation and tests are only on the master branch.
dhomm's avatar
dhomm committed
5
6

The Folder with thesis-code includes my result systems.
dhomm's avatar
dhomm committed
7
8
To run my systems the main branch should be used.

dhomm's avatar
dhomm committed
9
10
All systems except the edited imustat include the
properties for the Proband3.mkv file.
dhomm's avatar
dhomm committed
11
The presented results do not include the warnings of the read method from av.io.
dhomm's avatar
dhomm committed
12
13
14

If you clone the repo the Proband3.mkv file will be included.

dhomm's avatar
dhomm committed
15
To change the linted file, change the properties 
dhomm's avatar
dhomm committed
16
below the if \_\_name\_\_ == "\_\_main\_\_" clause accordingly.
dhomm's avatar
dhomm committed
17
18


dhomm's avatar
dhomm committed
19
20
PLOE-Adaption:
To run only the PLOE-Adaption go to the Wrist-Distinction
dhomm's avatar
dhomm committed
21
22
23
and run <br>
python3 PLOE_combined.py <br>
The result should then be: <br>
dhomm's avatar
dhomm committed
24
 > Using a timespan of:  0.9 and a threshold of:  (9, 3.5) the placement: 	 ('Position L1, X -> Elbow, Y -> Thumb, Z -> Palm', 1.0) was chosen
dhomm's avatar
dhomm committed
25
26
27



dhomm's avatar
dhomm committed
28
Placement Finder:<br>
dhomm's avatar
dhomm committed
29
To run only the Placement Finder for one file go to the placement_finder
dhomm's avatar
dhomm committed
30
31
32
and run<br>
python3 placement_finder.py<br>
The result should be:<br>
dhomm's avatar
dhomm committed
33
34
35
36
37
38
39
> For the metadata palcement:	 chest    	was the:  chest  	chosen<br>
> For the metadata palcement:	 forearm    was the:  forearm  	chosen<br>
> For the metadata palcement:	 head    	was the:  head  	chosen<br>
> For the metadata palcement:	 shin    	was the:  shin  	chosen<br>
> For the metadata palcement:	 thigh    	was the:  thigh  	chosen<br>
> For the metadata palcement:	 upperarm   was the:  upper arm chosen<br>
> For the metadata palcement:	 waist    	was the:  waist  	chosen<br>
dhomm's avatar
dhomm committed
40
41
42


Sensor Finder:<br>
dhomm's avatar
dhomm committed
43
To run only the Sensor Finder for one file go to the sensor_finder
dhomm's avatar
dhomm committed
44
45
46
and run:<br>
python3 sensor_finder.py<br>
The result should be:<br>
dhomm's avatar
dhomm committed
47
48
> For the wanted placement:  forearm the sensor with the metadata placement:  forearm was chosen.<br>
> All available metadata placements were:  ['chest', 'forearm', 'head', 'shin', 'thigh', 'upperarm', 'waist']<br>
dhomm's avatar
dhomm committed
49

dhomm's avatar
dhomm committed
50
Combined Sensor Finder:<br>
dhomm's avatar
dhomm committed
51
To run only the Sensor Finder for one file go to the combined-sensor-finder
dhomm's avatar
dhomm committed
52
53
54
and run:<br>
python3 combined-sensor-finder.py<br>
The result should be:<br>
dhomm's avatar
dhomm committed
55
56
> For the wanted placement:  forearm the sensor with the metadata placement:  forearm was chosen.<br>
> All available metadata placements were:  ['chest', 'forearm', 'head', 'shin', 'thigh', 'upperarm', 'waist']
dhomm's avatar
dhomm committed
57
58


dhomm's avatar
dhomm committed
59
The PLOE-Adaption and the Combined-Sensor Finder are also included in the imustat bei P. Scholl. <br>
dhomm's avatar
dhomm committed
60
This edited imustat is included in the edited-imustat folder. <br>
dhomm's avatar
dhomm committed
61
62
63
64
65
66
For details how to use the imustat check https://github.com/pscholl/imustat <br>
Additional to the original imustat two flags can be set: <br>
1. -w for wrist destinction <br>
Example: <br>
./imustat -w ../Thesis-Code/Proband3.mkv <br>
will result in: <br>
dhomm's avatar
dhomm committed
67
> Given Placement: forearm - PLOE-placement: (('Position L1, X -> Elbow, Y -> Thumb, Z -> Palm', 1.0), 0.9, (9, 3.5), 1) <br>
dhomm's avatar
dhomm committed
68
> \+ the result of the imustat <br>
dhomm's avatar
dhomm committed
69
70

2. -c for the combined sensor finder <br>
dhomm's avatar
dhomm committed
71
Example: <br>
dhomm's avatar
dhomm committed
72
73
./imustat -c ../Thesis-Code/Proband3.mkv <br>
will result in: <br>
dhomm's avatar
dhomm committed
74
75
> For the wanted placement:  wrist the sensor with the metadata placement:  forearm was chosen.<br>
> All available metadata placements were:  ['chest', 'forearm', 'head', 'shin', 'thigh', 'upperarm', 'waist']<br>
dhomm's avatar
dhomm committed
76
> \+ the result of the imustat <br>
dhomm's avatar
dhomm committed
77

dhomm's avatar
dhomm committed
78
Note: if -c is chosen during the process the wanted placement is asked for as input string. Please input this placement during the process.
dhomm's avatar
dhomm committed
79

dhomm's avatar
dhomm committed
80
81
<br>

dhomm's avatar
dhomm committed
82

dhomm's avatar
dhomm committed
83
84
85
All systems need the av.io module from https://github.com/pscholl/PyAV
all other requirements are easy installable with pip.

dhomm's avatar
dhomm committed
86