feature vector format

[WilliamWang1994]

download feature vector not match the description(angle,Δangle,Δpoint)

Hi i download the feature vector dataset from the link:https://drive.google.com/file/d/1ZNJDzQUjo2lDPwGoVkRLg77eA57dKUqx/view

the content of the download feature vector like this:

182.121,0.0372238,94.563,316.817,0.0580139,88.1386,76.9897,0.0069809,
85.9977,33.9192,0.0205917,80.3989,0,0,0,0,
0,0,0,0,0,0,0,0

182.137,0.540245,94.0227,316.774,0.898399,89.037,76.9441,0.00663757,
86.0043,36.2493,0.269905,80.6688,0,0,0,0,
0,0,0,0,0,0,0,0

i think the second line of the second feature vector must not the Δangle, so i check the all of dataset.i guess i nedd a description of the dataset feature vector format. thanks

[imsoo]

@WilliamWang1994

Thank you for your interest in this repo.

First of all, Sorry for the confusion.

The format of feature vector is as below. ( { Δpoint, Δangle, angle } Repeat 8 times )

IDX	1	2	3	4	5	6
Type	Δpoint	Δangle	angle	Δpoint	Δangle	angle
Joint	RELBOW	RSHOULDER - RELBOW	RSHOULDER - RELBOW	RWRIST	RELBOW - RWRIST	RELBOW - RWRIST
Value	182.121	0.0372238	94.563	316.817	0.0580139	88.1386

The joint data order is as below.

fight_detection/server/src/people.cpp

Lines 72 to 84 in f6a435c

	static const int change_part[] = {
	RELBOW, RWRIST, LELBOW, LWRIST, RKNEE, RANKLE, LKNEE, LANKLE
	};
	static const int change_pair[] = {
	RSHOULDER, RELBOW,
	RELBOW, RWRIST,
	LSHOULDER, LELBOW,
	LELBOW, LWRIST,
	RHIP, RKNEE,
	RKNEE, RANKLE,
	LHIP, LKNEE,
	LKNEE, LANKLE
	};

[WilliamWang1994]

@imsoo
i got it & very thankful.

[kenshinli2000]

@imsoo Hi, I was examining the dataset. I noticed that Δpoint of each vector of the same joint seems unusually large. Taking for example, pose3_stand.txt. The 1st element (corresponding to right elbow) of each feature vector for 1st 3 frames are :
182.121, ...
182.137, ..
179.913, ...

if Δpoint is 2D distance from previous position to the current position, shouldn't it be smaller and maybe even closer to zero for standing posture? and 180 pixels is quite a significant distance considering that the images are 720 x 480. Am I missing something?

Appreciate if you can help clarify. Thanks and great work on the project

[blackwool]

@imsoo Hi ,there
I have the same question of Δpoint value.
by the way ,are all the images size 720 * 480 ?
if not ,i think that normalize all the values to [0,1] will be better 。
Appreciate for you reply