哈喽,大家好。我是百变大魔王小明哥。对,没错,我就是传说中的文坛大佬、资深影评人、十八线歌手/演员。
##承接#各种男一号的戏##
哈哈,言归正传。
依旧记得某个时间我去百度大脑创新体验中心【在中关村创业大街,这里是创业公司一条街】,百度做的颜值评分是视频中抓图,速度比较慢,我有两次评测,88分/22岁、92分/24岁,可以,百度的实力还是很强的,前途不可限量啊,我很欣赏!!哈哈
因此有感而发,想做一下颜值评分?【好厉害哦】
寡人直接将75人[我仔细看了下是75人的评分数据]做平均,然后归一化到10分范围内,如果搞成百分制的,那么数据估计不够,那就误差太大了。本次数据来源:华工某人工智能课题组[公开数据],请直接搜即得。
据我看源码知:应该是做的分类,1~5分 5个类??
实际数据可能有些评分人的数据缺失了,如下:
本来500个图片,75个人评分,结果如上所示,少了一些。如果是这样的话,我可以做最坏打算,这里的顺序甚至都是错的,并不一定都是1,2,3,4这样的,但每一行的数据没有问题[这个有问题就没法玩了]。于是给出统计结果:归一化到0~10
array([ 3.838612 , 5.85084479, 3.41617211, 4.22557985, 3.50996185,
3.85332768, 4.38745231, 2.79379883, 4.88342518, 5.41013141,
3.95470236, 3.66256889, 4.34930055, 3.02597953, 2.47986435,
3.91600557, 2.92708775, 3.77702416, 5.03603222, 5.82397202,
4.46375583, 3.9296312 , 3.87730879, 3.99339914, 2.6003149 ,
4.49645734, 2.94858596, 4.19669351, 3.39550657, 3.47181009,
3.49034094, 4.65451462, 4.65451462, 5.0318512 , 7.08914189,
5.40291626, 6.66347726, 5.29726579, 7.70829044, 9.02183128,
4.76733483, 6.44189841, 4.9962908 , 3.85332768, 5.76091564,
8.0565615 , 7.24392903, 7.97916793, 6.94362018, 7.74480712,
6.16256889, 7.14692202, 6.86731666, 5.88954157, 3.91600557,
6.27650942, 7.97916793, 7.81793133, 4.22557985, 4.27299703,
3.6064313 , 3.41294737, 2.8995337 , 3.7171471 , 3.06467632,
3.6064313 , 3.85332768, 3.7388724 , 4.54005935, 6.23781263,
3.68382487, 4.96081875, 3.49034094, 3.33555381, 2.47986435,
4.76733483, 4.58245019, 5.23169624, 4.12039 , 3.87730879,
4.54005935, 4.45776055, 5.11560589, 4.10948949, 4.69266638,
3.37425059, 5.11560589, 4.8447284 , 5.38648338, 4.9962908 ,
3.838612 , 2.8995337 , 3.98050021, 3.98050021, 3.23460567,
4.10948949, 4.38745231, 4.80490997, 4.53010671, 5.22679101,
8.98891214, 9.11588597, 9.36983362, 9.36983362, 9.46163629,
8.56566607, 9.04196694, 8.8893599 , 6.23781263, 10. ,
8.52092291, 0.48166596, 3.4020639 , 1.05244353, 1.29715981,
5.22202204, 5.17969744, 5.22202204, 3.0902925 , 6.32246185,
3.52903773, 3.65601155, 5.812148 , 7.67957992, 5.42518016,
5.03391267, 4.61254769, 2.851844 , 3.41294737, 2.98728275,
4.26427663, 4.12039 , 9.53793981, 4.54005935, 5.19299945,
4.14818628, 5.68461212, 3.64512808, 4.18688306, 2.67770847,
3.77702416, 8.98528432, 4.96081875, 3.6365288 , 5.56061891,
4.49645734, 3.12844426, 3.49034094, 4.50190759, 3.05214074,
4.68994126, 3.33555381, 2.21334706, 0.93635318, 2.74692666,
2.71640526, 1.90758796, 4.38745231, 0. , 2.48422455,
2.51801611, 3.58626537, 3.34092836, 3.41294737, 3.31920305,
9.44964573, 8.62229758, 4.15854175, 9.84315388, 3.02597953,
3.50996185, 9.00381518, 3.34092836, 3.02597953, 1.98654091,
2.87119239, 8.48222612, 3.54811361, 5.38648338, 2.90988918,
3.91865604, 3.5666596 , 5.37939805, 2.29074063, 3.47181009,
9.36983362, 2.36275964, 8.90789075, 9.41094895, 2.51801611,
3.02597953, 2.55616787, 8.21134863, 8.94658754, 3.0023315 ,
8.43153879, 3.20474777, 3.16659602, 6.70217404, 3.05317187,
3.79991522, 3.89147944, 4.14818628, 2.17465028, 3.66256889,
2.21334706, 3.99339914, 2.56161812, 8.7918004 , 2.13649852,
4.05574396, 4.31114879, 3.96778296, 4.53515412, 3.39550657,
9.21746503, 5.65736087, 3.16659602, 2.56161812, 3.45164416,
2.83249561, 3.85332768, 2.98728275, 1.79313268, 3.54811361,
4.46375583, 4.76733483, 3.56773451, 3.49034094, 3.99339914,
2.09834676, 2.25095379, 4.18688306, 3.56773451, 5.34778659,
3.1033731 , 2.94858596, 2.94858596, 2.67062315, 4.72863804,
2.83249561, 7.10311573, 4.49645734, 3.68382487, 5.85084479,
4.19669351, 2.6003149 , 2.93768546, 3.11519712, 4.22557985,
3.39916496, 2.43439581, 1.78768243, 3.54759098, 3.39550657,
4.88342518, 2.88946588, 2.32943741, 3.49034094, 3.21946345,
2.48422455, 2.32943741, 3.39550657, 1.56422213, 3.58626537,
2.87119239, 1.23410343, 2.28910555, 2.2875159 , 3.64512808,
5.64646036, 2.82323018, 2.40038152, 2.71640526, 3.06467632,
4.13098771, 3.28105129, 5.7111064 , 3.56773451, 3.06467632,
1.64052565, 4.92212196, 1.83605341, 3.1033731 , 2.97583722,
2.28910555, 1.94246957, 2.6003149 , 2.94858596, 2.63901169,
3.14206988, 2.94858596, 4.68994126, 3.94287834, 4.9597287 ,
3.31920305, 3.29685702, 4.22557985, 5.64646036, 3.03995337,
2.17465028, 3.6064313 , 4.07079271, 3.49034094, 2.98728275,
4.84527342, 3.838612 , 3.79239084, 2.94858596, 4.07079271,
5.46387695, 5.0769091 , 2.75510204, 3.62441713, 4.26427663,
5.07418398, 2.05855992, 2.97583722, 4.22557985, 5.62555529,
1.94246957, 2.74692666, 3.20474777, 3.01398898, 4.04408648,
6.35390299, 3.25816024, 2.90988918, 2.66373463, 3.90525646,
2.98728275, 5.03603222, 1.94246957, 5.18440017, 4.04408648,
4.73081814, 2.47986435, 3.64512808, 1.55550173, 2.67770847,
4.58245019, 1.82637922, 4.2814752 , 4.30297341, 2.75510204,
4.03209592, 7.0199237 , 2.25095379, 5.15048749, 2.6003149 ,
3.28105129, 2.24989402, 4.22557985, 3.05214074, 3.99339914,
6.04432871, 4.8447284 , 4.66078615, 2.71640526, 3.81517592,
1.78768243, 3.86763459, 3.21946345, 5.03603222, 4.72863804,
4.15854175, 2.28910555, 2.67062315, 4.9597287 , 5.03821232,
2.1370284 , 3.39916496, 3.91600557, 2.29074063, 3.6064313 ,
3.07757524, 2.79379883, 4.8447284 , 3.9296312 , 6.04970326,
9.53793981, 9.76685036, 3.18076667, 2.93768546, 3.0902925 ,
2.39728931, 3.1033731 , 3.35735481, 6.14243323, 6.3152062 ,
5.95167444, 4.19669351, 5.23169624, 2.96470962, 3.70072064,
3.41294737, 2.8995337 , 3.1033731 , 5.46387695, 3.37425059,
2.28910555, 4.07079271, 2.29074063, 4.23484527, 1.74898565,
2.3681342 , 2.17465028, 2.87119239, 3.33555381, 1.24592745,
2.8995337 , 1.01374674, 3.1033731 , 1.67867741, 4.27299703,
4.10948949, 5.53200509, 2.47986435, 3.20474777, 3.05214074,
3.52903773, 3.85332768, 3.18076667, 4.80712166, 3.91600557,
4.46375583, 2.78507842, 2.59431963, 1.20723067, 2.13595349,
2.21280203, 1.94246957, 1.60237389, 2.13595349, 3.79991522,
2.3681342 , 4.30297341, 3.18076667, 3.34092836, 0.93312844,
1.87367529, 2.63901169, 3.66256889, 2.02204324, 4.2814752 ,
2.060195 , 2.82323018, 2.01986314, 4.58245019, 1.98116635,
3.96778296, 3.49034094, 3.77702416, 3.838612 , 2.8995337 ,
3.02597953, 2.32943741, 1.865076 , 2.13649852, 4.99788046,
2.8995337 , 3.33555381, 2.71640526, 2.66373463, 7.63089687,
8.25004542, 6.48579907, 6.86731666, 3.18076667, 3.76121843,
3.14206988, 2.7087749 , 5.22679101, 3.6064313 , 2.71640526,
2.17465028, 2.63901169, 2.67770847, 2.2875159 , 2.060195 ,
1.59419851, 3.95470236, 3.838612 , 4.65124447, 1.94891903,
2.78507842, 2.32943741, 5.61866408, 3.87730879, 2.97583722,
3.9296312 , 3.70072064, 5.68461212, 2.40683098, 4.76733483,
5.38648338, 3.29685702, 3.66256889, 4.73499916, 5.76091564])
我想先做个回归模型,然后再做个分类【1~10】
然而第一步都需要有人脸识别/人脸检测,这个库安装失败,我也是醉了
Windows下失败原因
服务器上是因为个人名下没空间了,我特么真的太难了。
后来搜索一下发现缺失cmake,按照这个大佬的博文安装cmake ok :注意,安装后关闭cmd重开
然后Windows下成功了
然而服务器咋安装呢??https://cmake.org/download/ 官网也不给个说明
参考一个博文:在无sudo权限下安装cmake,我添加的路径是:如下,切记一定按照人家给的安装。
vi ~/.bashrc
export PATH="/your/path/cmake-3.16.0-rc4/bin:$PATH"
最后一定要用source,不用则无效
【有想跑路的大佬,别删库了,把bashrc删了,环境就死了,试试如下:没有sudo权限就不加
sudo rm -rf ~/.bashrc
】
服务器也安装成功,
$ cmake
Usage
cmake [options] <path-to-source>
cmake [options] <path-to-existing-build>
cmake [options] -S <path-to-source> -B <path-to-build>
Specify a source directory to (re-)generate a build system for it in the
current working directory. Specify an existing build directory to
re-generate its build system.
Run 'cmake --help' for more information.
而目前我唯一能用的服务器在跑那个博文的东西,不能再跑项目了,只好Windows了。条件艰苦啊!
我试了下库的切图能力【人脸检测】还算一般,切得有点狠。
【图片数据来源于我的微信收藏:哈哈,本人保证绝不泄露任何人的隐私,除非得到允许和授权。我将给这些人脸打分,男的和丑的大可放心,你的照片我懒得看,哈哈。后期我将用这些数据进行人脸生成,到时候就可以公开GAN的数据了,哈哈,不知道有多少妹子要拉黑我了,晚了!![上面的纯属玩笑,数据源于facenet公开数据,不涉及隐私]】
本想直接参考别人的,结果face_detector这个库把我坑了,卧槽,注意啦,别安装这个库,它指定其他依赖库的版本为特定的某一个,而不是>=多少版本,这特么真是个大坑啊,吓得我赶紧终止了。卧槽,冷汗都出来了
还是自己画框吧。用Image搞得
下面我该做评分部分了,搞了半天才进入正题。emm是我太Low
但是我发现一个细节问题,有鬼??
图片命名格式是:如果检测到多个人脸,那么必有_1,_2来标记,这个特么的咋回事??少一个图被偷吃了???
而另外一个同样情况的则没有问题,卧槽
【后记:发现是draw的问题,但我一直没找到错在哪里,这是采用类的方式画框,算了,直接用改变数据的方式吧
错误提示如下:
Traceback (most recent call last):
File "<pyshell#172>", line 1, in <module>
draw.gty(f,outline=(0,28,0),width=3)
File "C:\.\AppData\Roaming\Python\Python36\site-packages\", line 247, in rectangle
ink, fill = self._getink(outline, fill)
File "C:\.\AppData\Roaming\Python\Python36\site-packages\", line 114, in _getink
ink = self.draw.draw_ink(ink)
TypeError: function takes exactly 1 argument (3 given)
】
关注细节问题事关成败,但对于性格来说也有好坏,很可能顾虑太多,投鼠忌器。说到底还是穷,有钱人才不会前怕狼后怕虎。
似乎人的影子也检测到了,这是人还是人的影子啊??
emm,下面这是俩人???你在逗我,看来这个库也不是100%啊
当然也有没找到 人脸的,而且是很明显的错误,一张图很明显的就是脸,emm,可能脸太大了,找不到。。。。。。
我已经设置,没找到脸的就不要画了,结果又出幺蛾子。。。。。。。。这是要气死我啊【后来发现是之前的数据没有删】
下面开始回归模型:
在对500张人脸识别时告诉我:搜了下说图像损坏不可用,不可用还能人脸识别??
UserWarning: Corrupt EXIF data. Expecting to read 2 bytes but only got 0.
在cmd下出错后再也不能存入新的图片了,最后只保存288个,而shell脚本下可以全部保存,尽管同样的错。
【仔细查看是因为标签名字相同,同样颜值的不少啊[看来想细化到100分不容易做啊,何况我这还是10分内]】
序号: 358
Warning (from warnings module):
File "C:\.\AppData\Roaming\Python\Python36\site-packages\PIL\TiffImagePlugin.py", line 802
warnings.warn(str(msg))
UserWarning: Corrupt EXIF data. Expecting to read 2 bytes but only got 0.
just warning, no balance for data preprocessing.So Let's begin.
Traing :referenced the keras version in github FaceRank
the result may be the best: val_loss 6.32,but the model is so big,187M-unbelievable !
will not wait for the end of the train if I find overfit in the train log.
Now will try the test for the biggest model,as follows
又出幺蛾子了,卧槽,分明将数据整到0~10之间了,结果测试下500的脸全部都在5之内,卧槽。咋回事鸭??!
这尼玛真是百思不得其解啊。卧槽
仔细查看了下6分以上的在500个中只有44个,数据严重不均衡。。。。。。。这样的话,能学到个鬼,高颜值的根本发觉不了。
怪我瞎整成10之内的,改变了数据分布。算了算了,寡人累了。
再一看,3.7分以上的只有45个左右,数据分布是正态分布真的好吗?尽管实际中的确如此,但我们见到的最多的还是4.5分以上的,因为分低的都视而不见自动忽略了,哈哈哈。
emm,这特么纯属扯淡了,模型有问题
predict score= 1.718684, true score= 1.197180
predict score= 1.808650, true score= 1.375000
predict score= 1.959868, true score= 1.541670
predict score= 2.240227, true score= 1.542860
predict score= 1.823932, true score= 1.571430
predict score= 1.960922, true score= 1.585710
predict score= 1.851269, true score= 1.642860
predict score= 2.441353, true score= 1.652780
predict score= 1.843677, true score= 1.657140
predict score= 2.771203, true score= 1.676060
predict score= 2.179285, true score= 1.771430
predict score= 2.411575, true score= 1.774650
predict score= 2.231192, true score= 1.785710
predict score= 1.513166, true score= 1.788730
predict score= 1.811773, true score= 1.802820
predict score= 1.877252, true score= 1.816900
predict score= 1.617169, true score= 1.842860
predict score= 2.452534, true score= 1.857140
predict score= 2.341800, true score= 1.857140
predict score= 2.113133, true score= 1.859150
predict score= 2.048269, true score= 1.871430
predict score= 2.107901, true score= 1.875000
predict score= 1.810346, true score= 1.885710
predict score= 1.915630, true score= 1.888890
predict score= 2.096509, true score= 1.901410
predict score= 2.048573, true score= 1.914290
predict score= 2.272068, true score= 1.914290
predict score= 2.167451, true score= 1.914290
predict score= 1.519081, true score= 1.914290
predict score= 1.902534, true score= 1.916670
predict score= 2.228856, true score= 1.928570
predict score= 2.199411, true score= 1.930560
predict score= 1.850452, true score= 1.942860
predict score= 1.911767, true score= 1.943660
predict score= 2.049870, true score= 1.957140
predict score= 1.869277, true score= 1.957750
predict score= 1.877480, true score= 1.957750
predict score= 2.493312, true score= 1.971830
predict score= 1.634227, true score= 1.985710
predict score= 1.811761, true score= 1.985710
predict score= 1.950510, true score= 1.985920
predict score= 2.259091, true score= 1.985920
predict score= 2.496687, true score= 1.986110
predict score= 1.610104, true score= 2.014080
predict score= 2.190854, true score= 2.014290
predict score= 2.067441, true score= 2.014290
predict score= 1.905296, true score= 2.027780
predict score= 1.821375, true score= 2.028170
predict score= 2.146346, true score= 2.028170
predict score= 2.219706, true score= 2.041670
predict score= 1.937982, true score= 2.041670
predict score= 2.181840, true score= 2.042250
predict score= 1.974387, true score= 2.042250
predict score= 2.087495, true score= 2.042250
predict score= 2.251927, true score= 2.042250
predict score= 2.645653, true score= 2.042860
predict score= 2.471311, true score= 2.042860
predict score= 2.032948, true score= 2.042860
predict score= 2.461302, true score= 2.057140
predict score= 1.996981, true score= 2.057140
predict score= 1.983116, true score= 2.057140
predict score= 2.173713, true score= 2.057140
predict score= 2.201285, true score= 2.069440
predict score= 2.085149, true score= 2.071430
predict score= 2.178454, true score= 2.071430
predict score= 1.834912, true score= 2.082190
predict score= 1.661496, true score= 2.083330
predict score= 2.214813, true score= 2.085710
predict score= 2.460681, true score= 2.095890
predict score= 1.819628, true score= 2.000000
predict score= 2.085342, true score= 2.000000
predict score= 1.746305, true score= 2.000000
predict score= 2.156060, true score= 2.000000
predict score= 2.156083, true score= 2.112680
predict score= 2.101058, true score= 2.112680
predict score= 1.855611, true score= 2.112680
predict score= 2.408434, true score= 2.112680
predict score= 1.880219, true score= 2.114290
predict score= 2.149660, true score= 2.114290
predict score= 1.975354, true score= 2.126760
predict score= 2.665725, true score= 2.126760
predict score= 2.529907, true score= 2.140850
predict score= 2.657847, true score= 2.142860
predict score= 2.154451, true score= 2.142860
predict score= 2.021443, true score= 2.154930
predict score= 2.487453, true score= 2.157140
predict score= 2.137426, true score= 2.157140
predict score= 1.805872, true score= 2.157140
predict score= 2.422641, true score= 2.157140
predict score= 2.226496, true score= 2.171430
predict score= 1.871087, true score= 2.171430
predict score= 2.179405, true score= 2.171430
predict score= 2.167445, true score= 2.180560
predict score= 2.127507, true score= 2.180560
predict score= 2.529450, true score= 2.183100
predict score= 2.364020, true score= 2.183100
predict score= 2.204085, true score= 2.185710
predict score= 2.202889, true score= 2.185710
predict score= 2.502658, true score= 2.185710
predict score= 2.298743, true score= 2.197180
predict score= 2.150085, true score= 2.211270
predict score= 2.725637, true score= 2.211270
predict score= 2.273650, true score= 2.214290
predict score= 2.269578, true score= 2.214290
predict score= 2.230801, true score= 2.225350
predict score= 2.098568, true score= 2.225350
predict score= 2.050913, true score= 2.228570
predict score= 2.439862, true score= 2.228570
predict score= 2.371387, true score= 2.239440
predict score= 2.055244, true score= 2.239440
predict score= 1.933375, true score= 2.242860
predict score= 2.384759, true score= 2.242860
predict score= 2.926850, true score= 2.257140
predict score= 2.464794, true score= 2.257140
predict score= 1.740831, true score= 2.257140
predict score= 2.156232, true score= 2.250000
predict score= 2.034206, true score= 2.263890
predict score= 2.075905, true score= 2.267610
predict score= 2.101728, true score= 2.267610
predict score= 1.618862, true score= 2.267610
predict score= 1.843977, true score= 2.267610
predict score= 2.245442, true score= 2.267610
predict score= 2.687862, true score= 2.267610
predict score= 2.064765, true score= 2.271430
predict score= 2.283896, true score= 2.271430
predict score= 2.254808, true score= 2.277780
predict score= 2.563015, true score= 2.281690
predict score= 1.904646, true score= 2.281690
predict score= 1.979742, true score= 2.285710
predict score= 2.492990, true score= 2.285710
predict score= 2.457470, true score= 2.285710
predict score= 1.977859, true score= 2.285710
predict score= 2.467931, true score= 2.285710
predict score= 1.500797, true score= 2.285710
predict score= 2.471922, true score= 2.291670
predict score= 2.048811, true score= 2.295770
predict score= 2.556995, true score= 2.295770
predict score= 2.346016, true score= 2.295770
predict score= 1.625766, true score= 2.200000
predict score= 2.429528, true score= 2.200000
predict score= 2.925484, true score= 2.200000
predict score= 2.221448, true score= 2.200000
predict score= 2.108144, true score= 2.200000
predict score= 2.295480, true score= 2.305560
predict score= 2.021484, true score= 2.309860
predict score= 2.369900, true score= 2.314290
predict score= 2.194779, true score= 2.314290
predict score= 1.973186, true score= 2.314290
predict score= 2.543079, true score= 2.314290
predict score= 1.825481, true score= 2.314290
predict score= 2.396766, true score= 2.319440
predict score= 2.021297, true score= 2.323940
predict score= 2.194350, true score= 2.323940
predict score= 2.071592, true score= 2.323940
predict score= 2.488394, true score= 2.324320
predict score= 2.467489, true score= 2.328570
predict score= 2.356105, true score= 2.328570
predict score= 2.295653, true score= 2.328570
predict score= 2.283711, true score= 2.333330
predict score= 1.969063, true score= 2.338030
predict score= 2.438116, true score= 2.338030
predict score= 1.973784, true score= 2.342860
predict score= 1.974746, true score= 2.342860
predict score= 2.521133, true score= 2.342860
predict score= 2.350390, true score= 2.342860
predict score= 1.565833, true score= 2.342860
predict score= 2.348210, true score= 2.347220
predict score= 2.464535, true score= 2.352110
predict score= 2.798338, true score= 2.357140
predict score= 2.575615, true score= 2.357140
predict score= 2.247730, true score= 2.366200
predict score= 1.997991, true score= 2.366200
predict score= 2.467346, true score= 2.371430
predict score= 1.631660, true score= 2.371430
predict score= 2.253988, true score= 2.371430
predict score= 2.105416, true score= 2.371430
predict score= 2.322992, true score= 2.380280
predict score= 2.094358, true score= 2.380280
predict score= 1.896040, true score= 2.380280
predict score= 2.892632, true score= 2.385710
predict score= 2.344193, true score= 2.385710
predict score= 2.323144, true score= 2.391300
predict score= 1.947908, true score= 2.300000
predict score= 2.382154, true score= 2.300000
predict score= 2.924040, true score= 2.300000
predict score= 2.262280, true score= 2.300000
predict score= 2.559019, true score= 2.408450
predict score= 2.355963, true score= 2.408450
predict score= 2.105036, true score= 2.414290
predict score= 2.012600, true score= 2.414290
predict score= 2.266146, true score= 2.422540
predict score= 2.678256, true score= 2.422540
predict score= 2.213278, true score= 2.428570
predict score= 1.728441, true score= 2.428570
predict score= 1.875831, true score= 2.428570
predict score= 2.258107, true score= 2.428570
predict score= 2.567168, true score= 2.430560
predict score= 2.167613, true score= 2.430560
predict score= 2.044671, true score= 2.430560
predict score= 2.724150, true score= 2.436620
predict score= 2.251495, true score= 2.442860
predict score= 2.518898, true score= 2.442860
predict score= 2.485991, true score= 2.450700
predict score= 2.121863, true score= 2.450700
predict score= 2.563817, true score= 2.450700
predict score= 2.184684, true score= 2.450700
predict score= 2.387010, true score= 2.452050
predict score= 1.868272, true score= 2.452050
predict score= 2.189674, true score= 2.453120
predict score= 2.014791, true score= 2.457140
predict score= 1.736987, true score= 2.457140
predict score= 2.317398, true score= 2.457140
predict score= 2.421174, true score= 2.457140
predict score= 2.128033, true score= 2.458330
predict score= 2.277461, true score= 2.471430
predict score= 1.924759, true score= 2.478870
predict score= 2.274218, true score= 2.478870
predict score= 2.459739, true score= 2.485710
predict score= 2.184052, true score= 2.485710
predict score= 2.340604, true score= 2.485710
predict score= 1.720737, true score= 2.485710
predict score= 2.629542, true score= 2.485710
predict score= 1.866092, true score= 2.485710
predict score= 2.438206, true score= 2.485710
predict score= 2.809671, true score= 2.492960
predict score= 2.154732, true score= 2.492960
predict score= 2.369350, true score= 2.400000
predict score= 2.867656, true score= 2.506850
predict score= 2.627836, true score= 2.507040
predict score= 2.485245, true score= 2.507040
predict score= 2.434034, true score= 2.513890
predict score= 2.191675, true score= 2.514290
predict score= 2.256542, true score= 2.514290
predict score= 2.447741, true score= 2.514290
predict score= 2.556174, true score= 2.521130
predict score= 2.432296, true score= 2.521130
predict score= 2.258201, true score= 2.528570
predict score= 2.266649, true score= 2.528570
predict score= 2.438804, true score= 2.528570
predict score= 2.414168, true score= 2.528570
predict score= 2.481726, true score= 2.528570
predict score= 2.492521, true score= 2.535210
predict score= 2.509201, true score= 2.539680
predict score= 1.892617, true score= 2.542860
predict score= 2.098646, true score= 2.542860
predict score= 2.451684, true score= 2.542860
predict score= 2.381890, true score= 2.546880
predict score= 2.279152, true score= 2.549300
predict score= 2.547939, true score= 2.549300
predict score= 1.901453, true score= 2.549300
predict score= 2.293876, true score= 2.549300
predict score= 2.283620, true score= 2.557140
predict score= 2.505839, true score= 2.557140
predict score= 2.531735, true score= 2.563380
predict score= 2.291978, true score= 2.563380
predict score= 2.497723, true score= 2.569440
predict score= 2.288044, true score= 2.577460
predict score= 2.594133, true score= 2.585710
predict score= 1.830582, true score= 2.591550
predict score= 2.602398, true score= 2.591550
predict score= 1.784458, true score= 2.591550
predict score= 2.423788, true score= 2.597220
predict score= 2.483742, true score= 2.500000
predict score= 1.989074, true score= 2.500000
predict score= 2.244470, true score= 2.605630
predict score= 2.380435, true score= 2.614290
predict score= 2.560905, true score= 2.614290
predict score= 1.579068, true score= 2.614290
predict score= 1.941065, true score= 2.614290
predict score= 2.605946, true score= 2.614290
predict score= 1.882363, true score= 2.619720
predict score= 2.171372, true score= 2.619720
predict score= 2.262565, true score= 2.619720
predict score= 2.525367, true score= 2.619720
predict score= 2.407974, true score= 2.619720
predict score= 2.196111, true score= 2.625000
predict score= 2.098053, true score= 2.628570
predict score= 2.163968, true score= 2.628570
predict score= 2.242512, true score= 2.628570
predict score= 2.280234, true score= 2.633800
predict score= 2.035549, true score= 2.638890
predict score= 2.550191, true score= 2.642860
predict score= 2.376678, true score= 2.642860
predict score= 2.117026, true score= 2.642860
predict score= 2.492067, true score= 2.642860
predict score= 2.503732, true score= 2.643840
predict score= 2.566395, true score= 2.647890
predict score= 2.463528, true score= 2.647890
predict score= 2.134969, true score= 2.647890
predict score= 2.331894, true score= 2.652780
predict score= 2.382891, true score= 2.657140
predict score= 2.036773, true score= 2.657140
predict score= 2.052912, true score= 2.661970
predict score= 2.181529, true score= 2.661970
predict score= 2.658096, true score= 2.666670
predict score= 2.248697, true score= 2.666670
predict score= 2.250623, true score= 2.671430
predict score= 2.620638, true score= 2.671430
predict score= 1.998400, true score= 2.671430
predict score= 2.408944, true score= 2.671430
predict score= 2.428775, true score= 2.685710
predict score= 2.445678, true score= 2.690140
predict score= 2.663245, true score= 2.690140
predict score= 1.713513, true score= 2.694440
predict score= 1.800315, true score= 2.600000
predict score= 2.025496, true score= 2.600000
predict score= 2.136575, true score= 2.714290
predict score= 2.359882, true score= 2.714290
predict score= 1.936338, true score= 2.714290
predict score= 2.384880, true score= 2.718310
predict score= 2.153433, true score= 2.718310
predict score= 2.151277, true score= 2.722220
predict score= 2.535037, true score= 2.728570
predict score= 2.201995, true score= 2.728570
predict score= 2.513853, true score= 2.732390
predict score= 2.221715, true score= 2.732390
predict score= 2.424844, true score= 2.742860
predict score= 2.363019, true score= 2.742860
predict score= 2.522990, true score= 2.746480
predict score= 2.643060, true score= 2.746480
predict score= 2.232424, true score= 2.746480
predict score= 2.089197, true score= 2.757140
predict score= 2.799828, true score= 2.757140
predict score= 1.999122, true score= 2.757140
predict score= 2.331923, true score= 2.757140
predict score= 2.152368, true score= 2.757140
predict score= 2.558501, true score= 2.757140
predict score= 1.922127, true score= 2.760560
predict score= 2.536332, true score= 2.771430
predict score= 2.337938, true score= 2.771430
predict score= 1.604730, true score= 2.774650
predict score= 2.334275, true score= 2.774650
predict score= 2.598979, true score= 2.777780
predict score= 2.230651, true score= 2.777780
predict score= 2.234789, true score= 2.785710
predict score= 2.443779, true score= 2.785710
predict score= 2.387771, true score= 2.788730
predict score= 2.445973, true score= 2.700000
predict score= 2.294485, true score= 2.700000
predict score= 1.707063, true score= 2.700000
predict score= 2.006861, true score= 2.802820
predict score= 1.794108, true score= 2.816900
predict score= 2.463207, true score= 2.816900
predict score= 2.347746, true score= 2.816900
predict score= 2.649271, true score= 2.842860
predict score= 2.168490, true score= 2.845070
predict score= 2.275237, true score= 2.845070
predict score= 2.254925, true score= 2.845070
predict score= 2.762258, true score= 2.857140
predict score= 2.008695, true score= 2.857140
predict score= 2.730464, true score= 2.857140
predict score= 2.691063, true score= 2.859150
predict score= 2.459685, true score= 2.869570
predict score= 2.719155, true score= 2.871430
predict score= 2.407809, true score= 2.873240
predict score= 2.565783, true score= 2.873240
predict score= 2.399129, true score= 2.873240
predict score= 2.727941, true score= 2.888890
predict score= 2.066723, true score= 2.888890
predict score= 2.252917, true score= 2.888890
predict score= 2.576528, true score= 2.914290
predict score= 2.636665, true score= 2.915490
predict score= 2.325417, true score= 2.915490
predict score= 2.367187, true score= 2.917810
predict score= 2.703972, true score= 2.928570
predict score= 2.312363, true score= 2.928570
predict score= 2.372347, true score= 2.929580
predict score= 2.675235, true score= 2.942860
predict score= 2.132493, true score= 2.942860
predict score= 2.486774, true score= 2.943660
predict score= 2.498966, true score= 2.945210
predict score= 2.126777, true score= 2.957140
predict score= 2.782847, true score= 2.957140
predict score= 2.219194, true score= 2.957140
predict score= 2.556586, true score= 2.957140
predict score= 2.074275, true score= 2.971010
predict score= 2.005307, true score= 2.971830
predict score= 2.536964, true score= 2.985710
predict score= 2.681552, true score= 2.985710
predict score= 2.186269, true score= 2.985710
predict score= 2.401629, true score= 2.985920
predict score= 2.295264, true score= 2.900000
predict score= 2.023592, true score= 3.014290
predict score= 2.502831, true score= 3.028170
predict score= 2.626818, true score= 3.028170
predict score= 2.056643, true score= 3.028570
predict score= 2.231221, true score= 3.028570
predict score= 2.386369, true score= 3.041670
predict score= 2.331984, true score= 3.041670
predict score= 2.110380, true score= 3.042250
predict score= 2.373262, true score= 3.054790
predict score= 2.673696, true score= 3.055560
predict score= 2.596211, true score= 3.056340
predict score= 2.196418, true score= 3.056340
predict score= 2.146620, true score= 3.056340
predict score= 2.489212, true score= 3.057140
predict score= 2.237127, true score= 3.070420
predict score= 2.238972, true score= 3.071430
predict score= 2.506298, true score= 3.085710
predict score= 2.477121, true score= 3.085710
predict score= 2.613854, true score= 3.098590
predict score= 2.050368, true score= 3.000000
predict score= 2.382693, true score= 3.000000
predict score= 2.157833, true score= 3.109380
predict score= 2.152849, true score= 3.111110
predict score= 2.739403, true score= 3.114290
predict score= 2.169423, true score= 3.125000
predict score= 2.757667, true score= 3.125000
predict score= 2.556656, true score= 3.126760
predict score= 2.526545, true score= 3.126760
predict score= 1.991197, true score= 3.128570
predict score= 2.549959, true score= 3.128570
predict score= 3.410070, true score= 3.152780
predict score= 2.704418, true score= 3.171430
predict score= 2.473015, true score= 3.183100
predict score= 2.688867, true score= 3.185710
predict score= 2.327588, true score= 3.185710
predict score= 2.387590, true score= 3.185710
predict score= 3.136008, true score= 3.191780
predict score= 2.359517, true score= 3.194440
predict score= 2.342927, true score= 3.214290
predict score= 2.290080, true score= 3.214290
predict score= 2.019578, true score= 3.239440
predict score= 2.208402, true score= 3.250000
predict score= 2.551875, true score= 3.271430
predict score= 2.061460, true score= 3.273970
predict score= 2.397108, true score= 3.281690
predict score= 2.423600, true score= 3.281690
predict score= 2.467963, true score= 3.285710
predict score= 2.682195, true score= 3.295770
predict score= 2.619075, true score= 3.295770
predict score= 2.649990, true score= 3.200000
predict score= 2.698799, true score= 3.305560
predict score= 3.079544, true score= 3.323940
predict score= 2.631431, true score= 3.323940
predict score= 2.534110, true score= 3.342860
predict score= 2.318895, true score= 3.347220
predict score= 2.513512, true score= 3.357140
predict score= 2.478343, true score= 3.357140
predict score= 2.785716, true score= 3.371430
predict score= 2.370877, true score= 3.394370
predict score= 2.857731, true score= 3.428570
predict score= 2.285085, true score= 3.430560
predict score= 2.704370, true score= 3.464790
predict score= 3.002744, true score= 3.472220
predict score= 2.852870, true score= 3.514290
predict score= 3.067591, true score= 3.528570
predict score= 2.655995, true score= 3.531250
predict score= 2.245549, true score= 3.542860
predict score= 2.491201, true score= 3.575340
predict score= 2.818178, true score= 3.591550
predict score= 2.240878, true score= 3.500000
predict score= 2.325336, true score= 3.500000
predict score= 2.582752, true score= 3.657140
predict score= 2.787440, true score= 3.671430
predict score= 3.211563, true score= 3.732390
predict score= 3.147078, true score= 3.732390
predict score= 3.027478, true score= 3.760560
predict score= 2.562897, true score= 3.788730
predict score= 3.267259, true score= 3.814290
predict score= 2.495403, true score= 3.819440
predict score= 2.933606, true score= 3.835620
predict score= 3.205050, true score= 3.871430
predict score= 2.598254, true score= 4.014290
predict score= 2.589359, true score= 4.032260
predict score= 2.817161, true score= 4.042860
predict score= 3.095720, true score= 4.056340
predict score= 2.969477, true score= 4.083330
predict score= 3.352556, true score= 4.142860
predict score= 3.226750, true score= 4.142860
predict score= 2.591966, true score= 4.171430
predict score= 2.788876, true score= 4.228570
predict score= 3.077783, true score= 4.242860
predict score= 3.218482, true score= 4.309860
predict score= 2.857835, true score= 4.328570
predict score= 3.235997, true score= 4.342860
predict score= 2.938557, true score= 4.359380
predict score= 2.962600, true score= 4.380280
predict score= 3.334735, true score= 4.442860
predict score= 2.866377, true score= 4.478870
predict score= 2.872341, true score= 4.485710
predict score= 3.430226, true score= 4.514290
predict score= 2.889666, true score= 4.515620
predict score= 2.851023, true score= 4.521130
predict score= 3.184268, true score= 4.527780
predict score= 2.788523, true score= 4.535210
predict score= 2.371465, true score= 4.562500
predict score= 3.263453, true score= 4.500000
predict score= 3.249276, true score= 4.656250
predict score= 3.184888, true score= 4.656250
predict score= 2.804224, true score= 4.656250
predict score= 3.041454, true score= 4.671430
predict score= 2.527597, true score= 4.685710
predict score= 3.136914, true score= 4.690140
predict score= 2.864124, true score= 4.600000
predict score= 2.583934, true score= 4.718310
predict score= 3.330996, true score= 4.718310
predict score= 3.463548, true score= 4.802820
predict score= 3.255983, true score= 4.830990
predict score= 3.161605, true score= 4.888890
直接测试500数据就是这狗样子,没法玩。回归的话数据还是不够。至少不够细化,所以说还是分类算了。
作为资深回归砖家,没有评价指标怎么可能,emm,你看看吧,我不想说话
超过50%误差的个数: 434
均方根误差: 3.1062065057541814
手动计算R^2————
R2=-3544.708161
sklearn计算r2————
r2=-5.942675
按道理说不能测试训练集,我连训练集都放进去了,这效果还是这么差劲,垃圾,模型垃圾,归根结底还是我垃圾。
请教了大神同事,修改了loss函数,结果有提高,但也不尽如人意,因为我的模型垃圾,这是关键。
超过20%误差的个数: 282
均方根误差: 0.774139995187755
手动计算R^2————
R2=-1033.273818
sklearn计算r2————
r2=-0.730278
r2如果是负的说明模型直接不能用。
我又加了block,与VGG差不多那种,结果如下,有提高,但仍旧不能用
超过20%误差的个数: 218
均方根误差: 0.5609995224490936
手动计算R^2————
R2=-867.115891
sklearn计算r2————
r2=-0.253889
【我将收藏的照片切割了一下,发现什么鬼都有,有表情包,有石像,这些我都忍了,你把狗头给我切出来是啥意思???啥头都不是的你也给我切出来了,我。。。。。
后来也发现画框的坐标有问题,如果是一个图多个人脸的话。
】
csdn is LowB, when I use My English answer the question,then delete by admin ??what ??stupid, foolish
寡人真的累了,搞了一天,同样的模型,结果我非要改参数,结果懵逼了,再也回不去了,以后一定要记住,每训练一次都要记住改了哪里。反正都不行,r2都是负数,不能用。上分类吧
先看看别人怎么标记的数据,不然又是瞎搞。数据不均衡太严重了。
限于篇幅及阅读的方便,请看下一篇:https://blog.csdn.net/SPESEG/article/details/103194376
另外有相关问题可以加入QQ群讨论,不设微信群
QQ群:868373192
语音深度学习群