0:00:17 | uh i think it is my pleasure to present my work here and then a |
---|
0:00:22 | nice young channel and um now working in france at C inside your uh the |
---|
0:00:27 | subject of my paper is kernel similarity is that active appearance models for face recognition |
---|
0:00:35 | uh first of all i'd beginning uh i want to use a few words um |
---|
0:00:41 | active appearance models which is the base of my work uh it is quite to |
---|
0:00:47 | the widely used to yin uh face recognition and the object tracking and sometimes for |
---|
0:00:54 | uh medical image processing uh they in sexual uh the in central idea of this |
---|
0:01:01 | matter is to uh build a model which contains pose the shape and texture information |
---|
0:01:08 | uh of the training data uh and then when we have a new face to |
---|
0:01:13 | recognise the model can generate a pen generates the |
---|
0:01:18 | appearance of the new phase of id three shows so uh but us using this |
---|
0:01:24 | matter that we can get the detail of the appearance of a new face and |
---|
0:01:30 | also the uh location of the landmarks on the ball boulder of the base and |
---|
0:01:36 | the dog runs uh so it is still quite powerful wow algorithm for sometimes for |
---|
0:01:44 | the uh space |
---|
0:01:47 | for the recognition of the face expressions except for uh and the this matter it |
---|
0:01:54 | has it's lame eight so one of the most the important to a problem is |
---|
0:02:01 | uh is that the quite sensitive to the illumination conditions uh that is so why |
---|
0:02:07 | we want to our improvement |
---|
0:02:11 | and in recent years there was quite a lot of researchers work on this problem |
---|
0:02:16 | the illumination condition and the uh there is some idea is so what are some |
---|
0:02:22 | of them the posted to at some action a some action no parametrise scene this |
---|
0:02:29 | model which is the red delay corresponding to the illumination and to decide he rates |
---|
0:02:36 | they depend it to this work by using this site here uh active appearance model |
---|
0:02:41 | is able to generate the face is in a quite dark the illumination or a |
---|
0:02:48 | very bright illumination but uh we know that the illumination that is not the worst |
---|
0:02:54 | thing for the illumination when the uh illumination come from one side of the face |
---|
0:02:59 | it to make that is half dark and half right and this is the more |
---|
0:03:04 | complicated thing and the this idea that in the work for this case and someone's |
---|
0:03:12 | supposed to uh apply a filter on the decoder for two can uh invariance condition |
---|
0:03:19 | uh for example uh for here a transformation all couple filters uh and this matter |
---|
0:03:29 | the also cost to lose some data use the information from the recognition images and |
---|
0:03:36 | there is a kind of quite a tradition idea is to use some other october |
---|
0:03:40 | no uh transformation instead of principal component and that is to extract the uh the |
---|
0:03:47 | most important variables uh in that database and the although it is quite old about |
---|
0:03:53 | so we believe a we find the and up operate the transformation uh it can |
---|
0:03:58 | work for this case and this is the initial idea of our work |
---|
0:04:06 | and in this page i want to presents the database aware working on it is |
---|
0:04:11 | called a cmu pose illumination and expression database of human faces and descent also in |
---|
0:04:19 | this data database is captured the in section and environment uh in the room there |
---|
0:04:25 | is several cameras are wrong that the volunteer and several flashes are around him and |
---|
0:04:31 | to make a different illumination conditions the flashes flash one by one to get the |
---|
0:04:39 | illumination and the here i give out some examples in this database uh for each |
---|
0:04:46 | person your uh us fourteen different pose of that is and for each post there |
---|
0:04:53 | is twenty different illumination conditions and we can see from these pictures some of the |
---|
0:04:58 | illumination it's quite complicated but complicated and heart to recognise |
---|
0:05:05 | one so we decided to use this database uh with data statistic and the light |
---|
0:05:10 | on the database uh here in this page the we show that histogram of the |
---|
0:05:16 | euclidean distance between each of actors uh of the uh each vectors O and so |
---|
0:05:24 | one is for here is the uh that terms of the shape and here that |
---|
0:05:29 | is the distribution of the texture vectors uh we can see that so for the |
---|
0:05:35 | shape vectors it is the close to a gaussian distribution and for that extra ones |
---|
0:05:40 | uh it is so quite beautiful passion distribution uh this result is quite interesting and |
---|
0:05:47 | the according to this result we decided to use the kernel to be able to |
---|
0:05:53 | a similarity |
---|
0:05:57 | uh similarity matrix the instead of the covariance matrix which is to use the in |
---|
0:06:03 | pca uh have to see that it's uh occur no it's not such a new |
---|
0:06:10 | id reading in this case requires the kernel pca came out to maybe twenty years |
---|
0:06:16 | ago and it is directly used to eating active shape models which is the a |
---|
0:06:22 | priori is work of active appearance model but that the alternatives and to continue to |
---|
0:06:28 | use it in active appearance model requires it is very complicated to uh reconstruct the |
---|
0:06:35 | phase and we construct that you major from the uh extraction features |
---|
0:06:42 | but uh for the active appearance model it is very important to reconstruct the images |
---|
0:06:48 | um here the proposed method we call it to a kernel similarity component analyse it |
---|
0:06:54 | is quite different from the kernel pca but uh sometimes it seems quite similar with |
---|
0:07:01 | each other and to mathematics star race is quite so it's clearly reading you might |
---|
0:07:08 | paper he i don't want to uh talk about the can uh mathematical conclusions uh |
---|
0:07:14 | just the procedure of this matter what uh is very simple which is to use |
---|
0:07:20 | the kernel oh |
---|
0:07:22 | we just use the kernel to build the uh similarity metrics and then calculates a |
---|
0:07:27 | the principal component to from the uh |
---|
0:07:31 | from the uh from up from the metrics it uh and then we get the |
---|
0:07:36 | eigen faces which uh which represents a to the most important variation in the database |
---|
0:07:45 | uh and here in this in this page what i want to this i want |
---|
0:07:50 | to show how the eigen faces a facts to the variation of the model on |
---|
0:07:58 | the left part is the uh result of the proposed the matter uh we can |
---|
0:08:04 | see that for the first and the for the third feature uh it is obviously |
---|
0:08:10 | control the illumination variance uh the illumination environments on the face but so for the |
---|
0:08:17 | principal component to analyze um |
---|
0:08:21 | the bar uh the variance is only between the genders are sometimes between the different |
---|
0:08:28 | to uh shape of the face |
---|
0:08:32 | so this result is tell us that so we have already choose the appropriate the |
---|
0:08:41 | transformation um and here is the experimental results on the right C is the a |
---|
0:08:48 | bit evaluation curves which what i don't like it's uh i like to see directly |
---|
0:08:54 | the |
---|
0:08:56 | the image is as we said before for uh yes the first column is the |
---|
0:09:01 | result of the proposed method and the in the middle column eight is the result |
---|
0:09:06 | of standard again and left column is the original you make use a which is |
---|
0:09:11 | a to recognize and a as we said uh standard and it works well when |
---|
0:09:20 | the illumination is not that complicated but so when it is how dark and half |
---|
0:09:26 | right is uh the a and that and work but so the proposed method it |
---|
0:09:31 | gives the quite good result |
---|
0:09:34 | um we also applied this method in the a rotation of the phase in this |
---|
0:09:41 | problem but the uh from the U matrix we can see that the improvement is |
---|
0:09:47 | not that of years uh only for some certain case it's some uh some change |
---|
0:09:54 | but not a lot |
---|
0:10:01 | uh and here is the conclusion the proposed kernel similarity is the active appearance models |
---|
0:10:06 | is robust to illumination and pose changes of the pc images with this signal them |
---|
0:10:12 | at their depict the fitting procedure can accurately thing sizes bases for my right to |
---|
0:10:18 | my dark affected by the illumination and say i have to emphasize that this method |
---|
0:10:25 | has the quite big limit that sits requests uh applied set accuracy at a alignment |
---|
0:10:32 | of the shape and texture vector if we couldn't do this it won't work would |
---|
0:10:37 | and so that the next step for our work is that so we want to |
---|
0:10:41 | make the matter to work on both the pose variation and illumination variations |
---|
0:10:48 | and that this all thank you |
---|
0:11:07 | estimate the parameter of the |
---|
0:11:20 | so i can understand you clustering |
---|
0:11:25 | yeah uh_huh |
---|
0:11:30 | is your rights yeah there is a |
---|
0:11:34 | i you mean you mean i think you mean this one yeah this one oh |
---|
0:11:39 | yeah this is also part of our work |
---|
0:11:42 | you see wow build a histogram of the uh of the mecc vectors and so |
---|
0:11:49 | we uh and to that sir it shows the gaussian distribution so we just use |
---|
0:11:54 | the uh |
---|
0:11:57 | variance here |
---|
0:12:00 | it is a portion and to the kernel we use is the portion so |
---|
0:12:07 | we think here that it that is it's represents the variance of the gaussian distribution |
---|
0:12:15 | so is that the |
---|
0:13:28 | uh excuse me that's what is the other uh algorithm called you see |
---|
0:13:39 | uh yes i think i heard that before |
---|
0:13:58 | yeah actually the principal common to analyze is quite close to the uh independence the |
---|
0:14:04 | common is except that that's a we try this matter what right do you depend |
---|
0:14:09 | and one that's it doesn't work quite good |
---|
0:14:29 | oh yeah |
---|
0:14:35 | uh_huh |
---|
0:14:37 | yeah actually what have uh convert opens up the uh ninety eight percent each of |
---|
0:14:44 | the information |
---|