AI与医学领域数据的结合早已是一个热门的方向，基于深度学习技术来开发辅助智能识别和检测模型对于疾病的高效智能化诊断有着重要的指导意义，这里本文的主要思想就是想要基于轻量级的CNN模型来尝试开发构建息肉识别系统，后续项目中会需要基于此项目来进一步开发构建检测端模型，首先看下实例效果，如下所示：
接下来看下数据集情况：
这里是基于mobilenet模型开发实现的息肉识别，首先来看下mobienet模型的实现：
def MobileNet(
input_shape=None, alpha=1.0, depth_multiplier=1, dropout=1e-3, classes=1000
):
img_input = Input(shape=input_shape)
x = convBlock(img_input, 32, alpha, strides=(2, 2))
x = dwConvBlock(x, 64, alpha, depth_multiplier, block_id=1)
x = dwConvBlock(x, 128, alpha, depth_multiplier, strides=(2, 2), block_id=2)
x = dwConvBlock(x, 128, alpha, depth_multiplier, block_id=3)
x = dwConvBlock(x, 256, alpha, depth_multiplier, strides=(2, 2), block_id=4)
x = dwConvBlock(x, 256, alpha, depth_multiplier, block_id=5)
x = dwConvBlock(x, 512, alpha, depth_multiplier, strides=(2, 2), block_id=6)
x = dwConvBlock(x, 512, alpha, depth_multiplier, block_id=7)
x = dwConvBlock(x, 512, alpha, depth_multiplier, block_id=8)
x = dwConvBlock(x, 512, alpha, depth_multiplier, block_id=9)
x = dwConvBlock(x, 512, alpha, depth_multiplier, block_id=10)
x = dwConvBlock(x, 512, alpha, depth_multiplier, block_id=11)
x = dwConvBlock(x, 1024, alpha, depth_multiplier, strides=(2, 2), block_id=12)
x = dwConvBlock(x, 1024, alpha, depth_multiplier, block_id=13)
x = GlobalAveragePooling2D()(x)
shape = (1, 1, int(1024 * alpha))
x = Reshape(shape, name="reshape_1")(x)
x = Dropout(dropout, name="dropout")(x)
x = Conv2D(classes, (1, 1), padding="same", name="conv_preds")(x)
x = Activation("softmax", name="act_softmax")(x)
x = Reshape((classes,), name="reshape_2")(x)
inputs = img_input
model = Model(inputs, x, name="mobilenet_%0.2f" % (alpha))
return model
MobileNet是一种轻量级的卷积神经网络模型，旨在在计算资源受限的移动设备上实现高效的图像分类和目标检测。其主要原理如下：
-
Depthwise Separable Convolution：MobileNet使用Depthwise Separable Convolution来减少参数量和计算量。这是一种将标准卷积分解成深度卷积（Depthwise Convolution）和逐点卷积（Pointwise Convolution）两个步骤的方法。深度卷积仅对输入的每个通道进行卷积，减少了卷积核的数量。逐点卷积使用1x1卷积核来将深度卷积的输出转化为期望的特征维度。这种分解有效降低了参数量，减少了计算量。
-
网络结构设计：MobileNet采用了基于深度可分离卷积的轻量网络结构。网络主要由一系列重复的卷积块和下采样层构成。卷积块包含了深度卷积、逐点卷积和激活函数。下采样层通常使用步长较大的深度可分离卷积来减少特征图的尺寸。通过这种设计，MobileNet减少了网络的深度和参数量，从而在较小的设备上实现了高效的推理。
优点：
- 轻量高效：MobileNet采用了Depthwise Separable Convolution和轻量网络结构，大大减少了参数量和计算量，使得它在计算资源受限的设备上运行速度快。
- 网络结构可定制：MobileNet的网络结构可以根据不同的需求和资源限制进行调整和定制。可以通过调整深度可分离卷积的层数和通道数来平衡准确性和模型大小。
缺点：
- 精度受限：由于网络结构的轻量化和参数减少，MobileNet相对于大型网络模型，如ResNet和Inception等，可能牺牲了一定的精度。
- 对复杂数据集的泛化能力有限：MobileNet在处理复杂数据集上的泛化能力可能相对较差，适用于较简单的图像分类和目标检测任务。
需要根据实际应用场景和资源限制来权衡使用MobileNet的优势和劣势。在资源受限的设备上，如移动设备或嵌入式系统，MobileNet是一种高效的选择，但在对准确性和复杂性要求较高的任务上，可能需要考虑更为复杂的网络结构。
部分图像数据编码错误，这里为了避免训练中断出现错误，事先过滤处理了，如下所示：
def filterImgs(dataDir="data/"):
"""
过滤无效图片
"""
for one_label in os.listdir(dataDir):
oneDir=dataDir+one_label+"/"
for one_pic in os.listdir(oneDir):
try:
one_img = Image.open(oneDir + one_pic)
one_img = one_img.convert('RGB')
except Exception as e:
print("Exception: ", e)
print("removing: ", oneDir + one_pic)
os.remove(oneDir + one_pic)
接下来是随机划分数据集，如下所示：
# 加载解析创建数据集
if not os.path.exists("dataset.json"):
train_dataset = []
test_dataset = []
all_dataset = []
classes_list = os.listdir(datasetDir)
classes_list.sort()
print("classes_list: ", classes_list)
with open("weights/classes.txt","w") as f:
for one_label in classes_list:
f.write(one_label.strip()+"\n")
print("classes file write success!")
num_classes=len(classes_list)
for one_label in os.listdir(datasetDir):
oneDir = datasetDir + one_label + "/"
for one_pic in os.listdir(oneDir):
one_path = oneDir + one_pic
try:
one_ind = classes_list.index(one_label)
all_dataset.append([one_ind, one_path])
except:
pass
train_ratio = 0.90
train_num = int(train_ratio * len(all_dataset))
all_inds = list(range(len(all_dataset)))
train_inds = random.sample(all_inds, train_num)
test_inds = [one for one in all_inds if one not in train_inds]
for one_ind in train_inds:
train_dataset.append(all_dataset[one_ind])
for one_ind in test_inds:
test_dataset.append(all_dataset[one_ind])
dataset = {}
dataset["train"] = train_dataset
dataset["test"] = test_dataset
with open("dataset.json", "w") as f:
f.write(json.dumps(dataset))
else:
with open("dataset.json") as f:
dataset = json.load(f)
train_dataset = dataset["train"]
test_dataset = dataset["test"]
with open("weights/classes.txt","r") as f:
classes_list=[one.strip() for one in f.readlines() if one.strip()]
print("classes_list: ", classes_list)
num_classes = len(classes_list)
print("train_dataset_size: ", len(train_dataset))
print("test_dataset_size: ", len(test_dataset))默认训练集-测试集比例为：9:1。
默认设定400次epoch的迭代计算，记录了训练过程中的loss和acc指标，如下所示：
0.5154747596153846 0.39809782608695654 1.2164945270006473 1.463877745296644
0.74609375 0.485054347826087 0.6819654452399566 1.5721689203511113
0.8022836538461539 0.40217391304347827 0.5187144568189979 1.8715582412222158
0.8312800480769231 0.44972826086956524 0.43426069140864104 2.1099254203879316
0.8467548076923077 0.38858695652173914 0.4049599589063571 2.324588900027068
0.8565204326923077 0.4375 0.37355997033703786 2.0007840654124385
0.8649338942307693 0.422554347826087 0.3477869880958818 2.169725459554921
0.8640324519230769 0.38315217391304346 0.35456895086771023 2.349976124970809
0.8644831730769231 0.53125 0.342520742247311 1.657330106134
0.8801081730769231 0.49728260869565216 0.3203347835761423 2.0639226540275244
0.8676382211538461 0.46195652173913043 0.3448066282300995 2.0199126316153486
0.8781550480769231 0.37228260869565216 0.31672593127363 2.674230181652567
0.8709435096153846 0.422554347826087 0.3340735392859922 2.081441982932713
0.8787560096153846 0.421195652173913 0.31750239360217863 2.3421299353889795
0.8753004807692307 0.41983695652173914 0.3376231001904951 2.512510569199272
0.8754507211538461 0.42527173913043476 0.3258774747642187 2.376193694446398
0.8715444711538461 0.4171195652173913 0.32688325070417845 2.3013603220815244
0.8814603365384616 0.5081521739130435 0.30448799279446787 2.0197541506394097
0.8816105769230769 0.4320652173913043 0.3044046882826548 2.317944365998973
0.8771033653846154 0.46059782608695654 0.32638321394244063 2.3432255931522534
0.8865685096153846 0.42527173913043476 0.30195715851508653 2.343923029692277
0.8781550480769231 0.34782608695652173 0.3149156108713494 2.805894618449004
0.8849158653846154 0.46603260869565216 0.29744300163852483 2.246819864148679
0.8796574519230769 0.42391304347826086 0.3122760229744017 2.5408245169598125
0.8843149038461539 0.41304347826086957 0.2980435914718188 2.9040553258812944
0.8844651442307693 0.44565217391304346 0.301070542468761 2.4935074059859565
0.8778545673076923 0.46875 0.31903693402329314 2.5308368672495303
0.8859675480769231 0.47690217391304346 0.2962351868358942 2.3352172322895215
0.8762019230769231 0.4171195652173913 0.3214864184578451 2.552992069202921
0.8817608173076923 0.3546195652173913 0.3080323341732415 3.0748863583025723
0.8859675480769231 0.4048913043478261 0.2998330883251933 2.8639659104139907
0.8780048076923077 0.3654891304347826 0.3131044839258091 2.6549240765364273
0.8813100961538461 0.4008152173913043 0.29940157852923643 2.437786770903546
0.8810096153846154 0.3858695652173913 0.3110712842227748 2.725678236588188
0.8819110576923077 0.39402173913043476 0.30586446364983344 2.745364329089289
0.8837139423076923 0.49728260869565216 0.29587585855132115 2.165059094843657
0.8853665865384616 0.41032608695652173 0.29879140187627995 2.6892030757406484
0.880859375 0.45516304347826086 0.30439808248327327 2.56989878675212
0.8861177884615384 0.4266304347826087 0.29264089558273554 2.4963120740392934
0.87890625 0.45652173913043476 0.30941993196924716 2.7461025974024897
0.8898737980769231 0.4076086956521739 0.2910010141607087 2.7433270008667656
0.8909254807692307 0.485054347826087 0.2936164183327212 2.269909195278002
0.8849158653846154 0.4375 0.2901983897810659 2.3913298378820005
0.8861177884615384 0.41032608695652173 0.30570005140124035 2.6114216213640957
0.8913762019230769 0.501358695652174 0.2792244608908032 2.255806109179621
0.8850661057692307 0.4633152173913043 0.29921287584763306 2.4893640694410903
0.8931790865384616 0.46875 0.28616888822916037 2.159032510674518
0.8871694711538461 0.43070652173913043 0.29577509704260874 2.753981243009153
0.8805588942307693 0.45652173913043476 0.3119760208870642 2.524707389914471
0.8834134615384616 0.46195652173913043 0.31217187227538 2.1665377046750938
0.8846153846153846 0.48097826086956524 0.30655765644489574 2.3538503646850586
0.8840144230769231 0.4701086956521739 0.30137708643451333 2.6392081250315127
0.8810096153846154 0.49728260869565216 0.3035102318972349 2.334437536156696
0.8832632211538461 0.4701086956521739 0.2975065018981695 2.7989701084468677
0.8919771634615384 0.44157608695652173 0.28025509832570183 2.729577733122784
0.8915264423076923 0.5203804347826086 0.2857905236216119 2.1341733466023984
0.8888221153846154 0.48233695652173914 0.28865195937956184 2.3932810918144556
0.8837139423076923 0.4116847826086957 0.3006662157937311 2.483801510023034
0.8880709134615384 0.41983695652173914 0.2955223592356421 2.7886516633241074
0.8891225961538461 0.42934782608695654 0.2838552057420692 3.0877410225246265
0.8817608173076923 0.39402173913043476 0.31288937635075015 3.0417406351669976
0.8820612980769231 0.46875 0.30851197733472174 2.3539296233135723
0.8810096153846154 0.501358695652174 0.30860623735218096 2.12857418993245
0.8858173076923077 0.49320652173913043 0.30020116351974696 2.0661076773767886
0.8880709134615384 0.3858695652173913 0.2871731912018731 2.768910314725793
0.8825120192307693 0.48097826086956524 0.2992731337614644 2.365106121353481
0.8870192307692307 0.40353260869565216 0.3022848378795271 3.2620466895725415
0.8853665865384616 0.5108695652173914 0.3016634952420226 2.230887700682101
0.8924278846153846 0.5258152173913043 0.28622997344399875 2.2626931356347124
0.8942307692307693 0.4796195652173913 0.27511951418665165 2.525853431743124
0.8820612980769231 0.39402173913043476 0.3043149922831127 2.827407816182012
0.888671875 0.4986413043478261 0.2855500362885113 2.48911336193914
0.8895733173076923 0.47282608695652173 0.2879279766124315 2.3041591696117236
0.8916766826923077 0.4945652173913043 0.2883217267584629 2.4073164566703467
0.884765625 0.5 0.3035451683144157 2.2481620363567187
0.8861177884615384 0.41983695652173914 0.29773898999421644 3.0533045633979468
0.8954326923076923 0.452445652173913 0.2754751418430645 2.847517200138258
0.8805588942307693 0.3491847826086957 0.30879578101806915 3.1635055956633193
0.8913762019230769 0.4320652173913043 0.2839208242053596 2.765966026679329
0.8849158653846154 0.41847826086956524 0.28731423686258495 2.853112210398135
0.8888221153846154 0.3858695652173913 0.2911150610754983 2.892658031505087
0.8946814903846154 0.4891304347826087 0.28125995732485676 2.299334660820339
0.8799579326923077 0.4891304347826087 0.30359778037438023 2.5078257218651148
0.8846153846153846 0.485054347826087 0.30712116156848 2.5113985175671787
0.8877704326923077 0.5230978260869565 0.28306188717341196 2.2123636473780093
0.8862680288461539 0.5258152173913043 0.2938474529207899 2.132711166920869
0.8874699519230769 0.47554347826086957 0.28447374888659954 2.502578263697417
0.8907752403846154 0.40353260869565216 0.2903130086532866 2.7296614595081494
0.8856670673076923 0.4633152173913043 0.29906475672928184 2.35868239402771
0.8951322115384616 0.5 0.27022388000757647 2.5580334767051367
0.8879206730769231 0.4429347826086957 0.28805010930563396 2.8339048986849575
0.8871694711538461 0.5108695652173914 0.29040914104105187 2.5183760031409888
0.8853665865384616 0.4782608695652174 0.29514566608346426 2.54559138028518
0.8852163461538461 0.4470108695652174 0.2896655301491802 2.843107969864555
0.8843149038461539 0.5353260869565217 0.2962305262649002 2.010954120884771
0.8897235576923077 0.4891304347826087 0.296751022983629 2.5319285703741987
0.8910757211538461 0.4986413043478261 0.2868757141001809 2.18648450270943
0.8883713942307693 0.53125 0.28492266045381814 2.0482969335887744
0.8946814903846154 0.422554347826087 0.27908764840461886 2.9815135935078496
0.8898737980769231 0.49320652173913043 0.2847877925870797 2.433269930922467
0.8877704326923077 0.421195652173913 0.2914668395398901 2.9109813026759936
0.8829627403846154 0.421195652173913 0.29618925826910597 2.9937719365824824
0.8921274038461539 0.36277173913043476 0.27185014001308727 2.7511361215425576
0.8880709134615384 0.4741847826086957 0.2914411407322265 2.38039491487586
0.8882211538461539 0.4578804347826087 0.2935656687029852 2.674441565638003
0.8838641826923077 0.47282608695652173 0.29322710331493557 2.443865242211715
0.8814603365384616 0.47554347826086957 0.3029458928638353 2.256706548773724
0.8904747596153846 0.36820652173913043 0.2842507206906493 3.283756266469541
0.8895733173076923 0.48097826086956524 0.2863835884222331 2.5577310220054956
0.8933293269230769 0.5 0.27664133453240186 2.285953506179478
0.8897235576923077 0.4633152173913043 0.28592746794367063 2.708962559700012
0.8859675480769231 0.43478260869565216 0.2881252178384994 3.0230617212212603
0.8868689903846154 0.43342391304347827 0.2907924360344903 2.6715903437655903
0.8895733173076923 0.483695652173913 0.28292465537714845 2.7363364748332812
0.8864182692307693 0.48097826086956524 0.287336961294596 2.4666915924652764
0.8901742788461539 0.53125 0.2798259420177111 2.1327932191931684
0.8831129807692307 0.4388586956521739 0.3002606089441822 3.0748048968937085
0.8909254807692307 0.5095108695652174 0.28100595016104096 2.447758068209109
0.8873197115384616 0.4470108695652174 0.29411856196104336 2.832267113353895
0.8840144230769231 0.44565217391304346 0.2924106385415563 2.7040689976319023
0.890625 0.4592391304347826 0.27290908159473193 2.8543530547100566
0.8849158653846154 0.4375 0.29191818297840655 2.6236291242682417
0.8862680288461539 0.421195652173913 0.30117998933061385 2.9916516283284063
0.888671875 0.4429347826086957 0.28470070398627567 2.809993168582087
0.8868689903846154 0.4741847826086957 0.2942093478587384 2.421338185020115
0.8889723557692307 0.3845108695652174 0.28771216556644785 3.2743787195371543
0.8841646634615384 0.3858695652173913 0.29487528580312544 2.8993696233500605
0.8939302884615384 0.45652173913043476 0.26520583897721595 2.826691472012064
0.8939302884615384 0.43342391304347827 0.2789954667767653 2.9413222851960557
0.8895733173076923 0.4429347826086957 0.2844223710350119 2.892515700796376
0.8864182692307693 0.5108695652173914 0.29274962837091434 2.7277851260226704
0.8960336538461539 0.45108695652173914 0.2670513882230108 2.921879706175431
0.892578125 0.4701086956521739 0.276901362499652 2.7721656301747197
0.8873197115384616 0.44972826086956524 0.2850624556438281 2.842447317164877
0.8910757211538461 0.5040760869565217 0.28625117471584904 2.332304633182028
0.8910757211538461 0.43478260869565216 0.2736904413725894 2.912581739218339
0.8913762019230769 0.4633152173913043 0.2878354856601128 2.8400660390439243
0.8967848557692307 0.46059782608695654 0.27548806473182946 2.6602804090665733
0.8892728365384616 0.44972826086956524 0.29633591300807893 2.950828033944835
0.8865685096153846 0.4701086956521739 0.28834600632007307 2.651406733886055
0.8892728365384616 0.5081521739130435 0.2977613198678367 2.547774542932925
0.8883713942307693 0.48641304347826086 0.27934261506351715 2.6460869675097256
0.8913762019230769 0.4578804347826087 0.2803412304761318 2.8015395662058955
0.8952824519230769 0.5095108695652174 0.28050794939582163 2.3571599929229072
0.8936298076923077 0.4361413043478261 0.2808823570286712 2.912127805792767
0.8943810096153846 0.46875 0.27543590319916034 2.870232965635217
0.8957331730769231 0.4388586956521739 0.27472519025636405 2.940248271693354
0.8949819711538461 0.44021739130434784 0.2686935191830763 3.029443419497946
0.8975360576923077 0.44157608695652173 0.2690277266710137 2.701536862746529
0.8864182692307693 0.46467391304347827 0.2826250663313728 2.6624635457992554
0.8951322115384616 0.5027173913043478 0.273179771235356 2.470261319823887
0.8882211538461539 0.47282608695652173 0.27425466776968765 2.6670112298882525
0.88671875 0.4945652173913043 0.29272530428492105 2.567967378574869
0.8910757211538461 0.5 0.281700994926863 2.4220072342001875
0.8882211538461539 0.4470108695652174 0.29281151889321894 2.823391644851021
0.8852163461538461 0.44429347826086957 0.28746507855920267 2.814295716907667
0.8924278846153846 0.46195652173913043 0.2758839978800657 3.002490893654201
0.8949819711538461 0.4782608695652174 0.2779910217146747 2.6343720322069912
0.8915264423076923 0.4361413043478261 0.28289966575371533 3.0601759267889936
0.8934795673076923 0.4673913043478261 0.267214122085044 2.678889995035918
0.8975360576923077 0.4904891304347826 0.26410165113898426 2.4440910712532373
0.8919771634615384 0.48777173913043476 0.27884331091235465 2.2524498545605205
0.8930288461538461 0.3532608695652174 0.27625790961946434 2.830244888430056
0.8955829326923077 0.41032608695652173 0.2755549062903111 2.962997395059337
0.8901742788461539 0.422554347826087 0.2871718997350679 2.7706573009490967
0.8951322115384616 0.38722826086956524 0.27266131336084354 3.0875673501387886
0.8871694711538461 0.47282608695652173 0.293385399863697 2.806963993155438
0.8928786057692307 0.42527173913043476 0.28919891885911614 3.0668149098106054
0.90234375 0.4741847826086957 0.2630252827388736 2.6485732482827227
0.8991887019230769 0.39809782608695654 0.26983211914865446 3.012707233428955
0.8948317307692307 0.4782608695652174 0.2672119884250256 2.436317547507908
0.8879206730769231 0.44157608695652173 0.29765915476645416 2.81890957251839
0.8928786057692307 0.47690217391304346 0.26492353338891494 2.4909602559131123
0.8982872596153846 0.42527173913043476 0.2733858806045296 2.657066936078279
0.8907752403846154 0.43070652173913043 0.28357432449523073 2.9273608611977617
0.8928786057692307 0.4741847826086957 0.2863177219584871 2.480356164600538
0.8928786057692307 0.46195652173913043 0.2823776298363765 2.570308540178382
0.8918269230769231 0.4429347826086957 0.2798408408505985 2.6541925357735674
0.892578125 0.45516304347826086 0.28498500341979355 2.7680121660232544
0.8907752403846154 0.47282608695652173 0.27438178355805576 2.577514560326286
0.8946814903846154 0.4483695652173913 0.2717607286042319 2.872758502545564
0.8973858173076923 0.48777173913043476 0.26058577955700457 2.555363178253174
0.8982872596153846 0.41983695652173914 0.2668059213946645 2.9633945174839185
0.8942307692307693 0.47146739130434784 0.2632574850979906 2.671549714129904
0.8957331730769231 0.43342391304347827 0.2704022080064393 2.800240910571554
0.8895733173076923 0.4388586956521739 0.28557478117111784 2.751232085020646
0.89453125 0.41032608695652173 0.27291692295469916 2.9171371511791064
0.8936298076923077 0.43478260869565216 0.27177835280935353 2.622857653576395
0.8913762019230769 0.46195652173913043 0.2760547229196303 2.7482729424601016
0.8973858173076923 0.4673913043478261 0.27051021287647575 2.687162731004798
0.8948317307692307 0.452445652173913 0.27357070445298004 2.731265394583992
0.888671875 0.4701086956521739 0.28516063685170734 2.671067388161369
0.8919771634615384 0.38179347826086957 0.27789646082629377 2.9050177854040395
0.8903245192307693 0.4578804347826087 0.27724158339417326 2.657291484915692
0.8933293269230769 0.5095108695652174 0.27831705978426796 2.421064511589382
0.8898737980769231 0.4375 0.28837586029504353 2.7983472761900527
0.9008413461538461 0.46195652173913043 0.25650350174580056 2.7743550279866094
0.8990384615384616 0.4904891304347826 0.26886529684998095 2.443142662877622
0.8985877403846154 0.485054347826087 0.2660124377311709 2.703730417334515
0.8912259615384616 0.4986413043478261 0.2697861988025789 2.3042280362999956
0.8598257211538461 0.7432065217391305 0.4212328173005237 1.015492866868558
0.9070012019230769 0.5543478260869565 0.269506339288245 3.278606839801954
0.9256310096153846 0.9660326086956522 0.1997622530683517 0.1067177899910704
0.9326923076923077 0.8926630434782609 0.17713574500969395 0.27043487439336983
0.9388521634615384 0.8695652173913043 0.16880519271720773 0.2978472427829452
0.939453125 0.873641304347826 0.1560389738700066 0.40900690529657446
0.9409555288461539 0.9456521739130435 0.1581013178309569 0.16054296485431815
0.9430588942307693 0.9592391304347826 0.1424108164725252 0.1725385525547292
0.9456129807692307 0.9619565217391305 0.13401868186282137 0.11228455335873624
0.947265625 0.7744565217391305 0.13112430714178258 1.1612390823986218
0.9471153846153846 0.9578804347826086 0.12766221242670256 0.11394871836123259
0.9489182692307693 0.9605978260869565 0.12287751093166523 0.11783750429648258
0.953125 0.7540760869565217 0.11246333584131207 1.199343696884487
0.9538762019230769 0.9429347826086957 0.11620759786232017 0.20155683678129446
0.9538762019230769 0.6820652173913043 0.11088700519758277 1.9666974622270335
0.9532752403846154 0.9592391304347826 0.11604829098527822 0.11509012370167868
0.9555288461538461 0.9619565217391305 0.10896476781636011 0.1059316041190987
0.9541766826923077 0.8288043478260869 0.10589474954534896 0.6907745249893354
0.9549278846153846 0.9592391304347826 0.10781335283984216 0.13076722735296126
0.9562800480769231 0.9429347826086957 0.10295770003442438 0.12272231418477453
0.9579326923076923 0.9497282608695652 0.10230837394080411 0.15555601615620696
0.9573317307692307 0.9171195652173914 0.09916634050806841 0.1521668275413306
0.95703125 0.9619565217391305 0.09746711085944508 0.11852377802943406
0.9568810096153846 0.7214673913043478 0.10240870227482027 1.3144661939662436
0.9585336538461539 0.9633152173913043 0.09885885442333976 0.11288989088295595
0.9568810096153846 0.9578804347826086 0.09724744179682322 0.11856496617521929
0.9582331730769231 0.9619565217391305 0.0917895897582639 0.11756004911401997
0.9552283653846154 0.9646739130434783 0.09890473421322755 0.11319699170037537
0.9601862980769231 0.9402173913043478 0.08759136572301102 0.13894028734901678
0.9585336538461539 0.9252717391304348 0.09484237361287412 0.14958441856762636
0.9627403846153846 0.9307065217391305 0.08853643841575831 0.14436438598710558
0.9595853365384616 0.9646739130434783 0.08759786136215553 0.10861176513540356
0.9627403846153846 0.9510869565217391 0.08897888747966275 0.13516570059785052
0.9618389423076923 0.8831521739130435 0.08706979445280293 0.4086050851189572
0.9612379807692307 0.9130434782608695 0.08969144215314792 0.1690959068744079
0.9606370192307693 0.9619565217391305 0.08747194374266726 0.10812375344255049
0.9624399038461539 0.9293478260869565 0.08368697476376277 0.15371871950185817
0.9582331730769231 0.7350543478260869 0.09069135622737051 1.393590763859127
0.9621394230769231 0.9497282608695652 0.08263452409300953 0.12613236418236856
0.9651442307692307 0.9415760869565217 0.08160244606551714 0.1380750992861779
0.9637920673076923 0.9510869565217391 0.08454167486231129 0.15291154200373136
0.9616887019230769 0.9578804347826086 0.08483155266954921 0.13714137027526033
0.9612379807692307 0.9755434782608695 0.0834715349617629 0.10102907286794938
0.9630408653846154 0.9524456521739131 0.0807554003997491 0.13634527183097342
0.9625901442307693 0.9565217391304348 0.07921249677803224 0.13621016350858237
0.9643930288461539 0.9551630434782609 0.0776787821904201 0.1563793986266398
0.9627403846153846 0.938858695652174 0.07884910295027882 0.15982770421744688
0.9673978365384616 0.9415760869565217 0.07452325086789922 0.14051217160633075
0.9637920673076923 0.936141304347826 0.0790394495259254 0.15583667282581978
0.9661959134615384 0.8586956521739131 0.07550027928650357 0.47083679003560025
0.9666466346153846 0.9402173913043478 0.07790034638870005 0.1441653453461502
0.9649939903846154 0.9320652173913043 0.07327305444946083 0.17035680782535803
0.96484375 0.936141304347826 0.07614161825935858 0.14239099296822172
0.9651442307692307 0.9402173913043478 0.07680537250192944 0.13579100224634874
0.9699519230769231 0.9470108695652174 0.06933153681396877 0.1438648452091476
0.9663461538461539 0.9510869565217391 0.07276850945345359 0.1354364890443242
0.9722055288461539 0.9551630434782609 0.06706397987503666 0.14891471495123013
0.9655949519230769 0.9497282608695652 0.07768085598945618 0.14486782823729774
0.9679987980769231 0.9538043478260869 0.06987947401536915 0.15535697114208472
0.9711538461538461 0.9334239130434783 0.06603014429744619 0.16611284829433198
0.9710036057692307 0.9592391304347826 0.06561606363370978 0.1312575338448843
0.9696514423076923 0.9375 0.06771826268959558 0.17458214473141276
0.9679987980769231 0.9497282608695652 0.07175122190388859 0.14467759900118993
0.9716045673076923 0.9415760869565217 0.0603583000467132 0.17769713505454685
0.9701021634615384 0.9456521739130435 0.06915450326834197 0.1427680142223835
0.9692007211538461 0.9442934782608695 0.07002855833917028 0.1594379726878327
0.9699519230769231 0.9497282608695652 0.06675829201524791 0.1452396148815751
0.9699519230769231 0.9307065217391305 0.06727125447762844 0.1884279001666152
0.97265625 0.9375 0.06359289310281523 0.14785364472671694
0.9710036057692307 0.9442934782608695 0.06528271435056993 0.16405626767031525
0.9719050480769231 0.7730978260869565 0.05937542561710311 0.9209518678810286
0.9696514423076923 0.9456521739130435 0.06538690295509206 0.16750349711788737
0.9740084134615384 0.9266304347826086 0.061936326996566586 0.17650983630396103
0.9675480769230769 0.9320652173913043 0.06448002191161951 0.1901563728309196
0.97265625 0.9334239130434783 0.061051013053936634 0.14732481304394163
0.9719050480769231 0.9456521739130435 0.060758976225490466 0.18843124596321065
0.9743088942307693 0.9510869565217391 0.058232051533895386 0.156291724982388
0.9708533653846154 0.9551630434782609 0.06508539441934912 0.13943104640297269
0.9714543269230769 0.9483695652173914 0.06106703921418207 0.15019657458786084
0.970703125 0.9415760869565217 0.06211045496568728 0.18714634273403688
0.9702524038461539 0.9429347826086957 0.06069733662969576 0.17426032512246267
0.9735576923076923 0.9605978260869565 0.06223451679957529 0.15214428807603722
0.9734074519230769 0.9592391304347826 0.05859835765470332 0.1387298446472572
0.9728064903846154 0.9483695652173914 0.05963121925651649 0.20609334101090612
0.9735576923076923 0.9565217391304348 0.06061598067865886 0.13530116288861987
0.97265625 0.9470108695652174 0.05756676650064317 0.17539020681389325
0.9740084134615384 0.9538043478260869 0.056590009039679826 0.17072418349309135
0.9779146634615384 0.9592391304347826 0.05367860949403481 0.13717470503331203
0.9752103365384616 0.9402173913043478 0.0577723619801243 0.18357371405252945
0.9735576923076923 0.9456521739130435 0.05724632404095386 0.18338390261582707
0.9744591346153846 0.9442934782608695 0.05670346738434791 0.19027767565263354
0.9771634615384616 0.9456521739130435 0.049958568283238074 0.17598935127582238
0.9798677884615384 0.9524456521739131 0.0477093883796792 0.18833276443183422
0.9773137019230769 0.9470108695652174 0.05544665716913117 0.1834350163521974
0.9728064903846154 0.9497282608695652 0.05782148010341469 0.16307297976606566
0.9759615384615384 0.9470108695652174 0.05428400979984032 0.18126220923200573
0.9759615384615384 0.9578804347826086 0.05345431073846367 0.17642718176964833
0.9737079326923077 0.9402173913043478 0.05747592884486389 0.18377926022462224
0.9777644230769231 0.9442934782608695 0.05193322618582057 0.1918139983292507
0.9770132211538461 0.9497282608695652 0.05182896202649527 0.17220531730487457
0.9791165865384616 0.9334239130434783 0.05054853064351706 0.19391078161804573
0.9783653846153846 0.9415760869565217 0.04850271838054491 0.1815838515556053
0.9722055288461539 0.9497282608695652 0.05920970412821939 0.18456623626306004
0.9803185096153846 0.936141304347826 0.04561567636630151 0.21254644544957124
0.9767127403846154 0.9538043478260869 0.049355140853619486 0.17224647691878286
0.9750600961538461 0.9320652173913043 0.05489308323012665 0.18301702444644077
0.9792668269230769 0.9578804347826086 0.04884947778044499 0.14222566637655962
0.9765625 0.938858695652174 0.052177301541423034 0.21136548159563023
0.9789663461538461 0.9470108695652174 0.048863304984339405 0.1824850377869671
0.9770132211538461 0.9538043478260869 0.052307245218029354 0.187277604517041
0.9768629807692307 0.9239130434782609 0.05057695943217438 0.21569863156132077
0.9777644230769231 0.9375 0.05003865255387679 0.2065663603539376
0.978515625 0.9456521739130435 0.04859285074976819 0.15998649058620568
0.9788161057692307 0.9442934782608695 0.04855424452342469 0.2313597056326335
0.9771634615384616 0.9470108695652174 0.05040304333628088 0.1795335964947615
0.9770132211538461 0.9456521739130435 0.052662830037521556 0.20274742711942806
0.9779146634615384 0.9402173913043478 0.049361286680393204 0.17622049991041422
0.9773137019230769 0.9483695652173914 0.04617622892412608 0.18444545946777394
0.9797175480769231 0.9415760869565217 0.044571893927870684 0.18674243541191454
0.9791165865384616 0.9470108695652174 0.044213062519422516 0.19548630649628845
0.9800180288461539 0.9497282608695652 0.04497824175912403 0.19020815212886943
0.9794170673076923 0.938858695652174 0.04467525872090706 0.2051856924334298
0.9767127403846154 0.9483695652173914 0.05103650511199913 0.1996752929509334
0.9779146634615384 0.9415760869565217 0.04662909196513983 0.1985785875307477
0.9794170673076923 0.936141304347826 0.04930423702367429 0.1937180313322207
0.9806189903846154 0.9429347826086957 0.04402356656926433 0.20036879558440135
0.9813701923076923 0.9524456521739131 0.04425908683949204 0.15839096758028734
0.9800180288461539 0.9456521739130435 0.0469821347053137 0.21845647968787132
0.9807692307692307 0.9470108695652174 0.041054458930515327 0.22594231296249706
0.9770132211538461 0.9375 0.05193418952638659 0.15642295964062214
0.9800180288461539 0.9334239130434783 0.04324858545078314 0.24183543463763985
0.9803185096153846 0.9470108695652174 0.04401226919547368 0.180644297759737
0.9812199519230769 0.9402173913043478 0.041690107198799245 0.20068481098860502
0.9806189903846154 0.9375 0.04432743249176733 0.19824959021871505
0.9801682692307693 0.9347826086956522 0.04190609308016764 0.23334059418867464
0.9815204326923077 0.9565217391304348 0.04403424169812262 0.17446431189613498
0.9791165865384616 0.9442934782608695 0.0447500769121409 0.18581873158235912
0.978515625 0.9402173913043478 0.04743666651875873 0.20852652366232613
0.9812199519230769 0.9442934782608695 0.04298889765498014 0.17858407952134375
0.9786658653846154 0.936141304347826 0.04355602167141534 0.22165330454869114
0.9819711538461539 0.9470108695652174 0.04178643593555675 0.19398524379357696
0.9795673076923077 0.9347826086956522 0.04367016721348941 0.22026323571639217
0.9809194711538461 0.9429347826086957 0.04455814160162565 0.19403020757042966
0.9816706730769231 0.9456521739130435 0.04125924324118667 0.20019776390298552
0.9813701923076923 0.9456521739130435 0.04289776419976148 0.20082146085231847
0.9788161057692307 0.9456521739130435 0.04649309000636164 0.1923993982212699
0.9809194711538461 0.9456521739130435 0.04260789693720513 0.223775469009643
0.9810697115384616 0.9402173913043478 0.0425325034753209 0.2083872464718297
0.9827223557692307 0.9429347826086957 0.04026543066006875 0.20150816497271476
0.9807692307692307 0.9497282608695652 0.04362037079324811 0.18870975927490255
0.9801682692307693 0.9483695652173914 0.04324448031792809 0.18737525739909514
0.9818209134615384 0.9320652173913043 0.04328284093823571 0.212721022443198
0.9819711538461539 0.936141304347826 0.04251168659105203 0.2513317824543818
0.9809194711538461 0.9429347826086957 0.04495084425042688 0.19596129906890186
0.9815204326923077 0.9456521739130435 0.043353336582815524 0.17945299039939014
0.9834735576923077 0.9293478260869565 0.037332745909225196 0.259054579043194
0.9806189903846154 0.9524456521739131 0.0433379349801376 0.1658448082718836
0.9815204326923077 0.9442934782608695 0.04047241779699107 0.22549053322037924
0.9809194711538461 0.9415760869565217 0.04345633311095298 0.2273846987594405
0.98046875 0.9483695652173914 0.042025313003530476 0.16071579051102794
0.9791165865384616 0.9470108695652174 0.04326277513777873 0.17884452932554742
0.9798677884615384 0.936141304347826 0.043055135188926494 0.21178825636920723
0.9809194711538461 0.9524456521739131 0.04118108507534019 0.2028218840562698
0.9774639423076923 0.9375 0.04753454151516333 0.22450798252880896
0.9819711538461539 0.9497282608695652 0.040747074870383725 0.1807518426541482
0.9800180288461539 0.9442934782608695 0.042177971253225296 0.2125108668418682
0.9831730769230769 0.9456521739130435 0.04076371839955055 0.1964559689083177
0.9819711538461539 0.9429347826086957 0.038143587275744024 0.2036857379235975
0.9813701923076923 0.9456521739130435 0.042731816436571535 0.20527269436127465
0.9813701923076923 0.9456521739130435 0.04047032824326799 0.21024836554779144
0.9807692307692307 0.938858695652174 0.04338226540354439 0.21516707250797798
0.9836237980769231 0.9415760869565217 0.03955918669446244 0.21384809529611273
0.978515625 0.9510869565217391 0.043211904333405146 0.18551778859134926
0.9834735576923077 0.9483695652173914 0.03824753260106753 0.18827630799395315
0.9822716346153846 0.9497282608695652 0.04040624508930845 0.1778528825379908
0.98046875 0.9347826086956522 0.041367857818841 0.24180647698433502
0.9836237980769231 0.9456521739130435 0.03946849920377925 0.21254267458520507
0.9818209134615384 0.9483695652173914 0.038994924209943674 0.19110952352133134
0.9812199519230769 0.9497282608695652 0.04163269028112364 0.1977178234121074
0.9807692307692307 0.9470108695652174 0.04097954266035231 0.18991075441970126
0.9797175480769231 0.9470108695652174 0.04592188870251098 0.18352113757282495
0.9806189903846154 0.9402173913043478 0.043144331519775175 0.2192580147150337
0.9819711538461539 0.9456521739130435 0.039872977852260544 0.20926870861212196
0.9806189903846154 0.9483695652173914 0.03887190216244982 0.1797277572483796
0.9810697115384616 0.9470108695652174 0.042583192022553146 0.20250466164282482
0.9818209134615384 0.9429347826086957 0.038324670578171766 0.20854254543740788
0.982421875 0.9497282608695652 0.03988815995175033 0.19960044534957927
0.9828725961538461 0.9429347826086957 0.04008368882715821 0.22782816130505956
0.9812199519230769 0.9524456521739131 0.040091900089884384 0.17747803048595137
0.9852764423076923 0.9442934782608695 0.03662994625035655 0.21332521456276046
0.9827223557692307 0.9483695652173914 0.042593549519481785 0.19427428904758848
0.9825721153846154 0.9483695652173914 0.03751505648369857 0.2034865390463813
0.9818209134615384 0.9470108695652174 0.03944827117028091 0.20417926517193732
0.9831730769230769 0.9524456521739131 0.03950558457626567 0.19233236537294704
0.9807692307692307 0.9415760869565217 0.040359040952423056 0.2154917812693621
0.9806189903846154 0.9442934782608695 0.042878100665872734 0.21596553931822596
0.9816706730769231 0.9538043478260869 0.0380077968436056 0.19643985715937679
0.9819711538461539 0.9470108695652174 0.03952874999553247 0.21059925655034653
0.9810697115384616 0.9524456521739131 0.04102906323029087 0.19199126809025588
0.9816706730769231 0.9510869565217391 0.040327877668600055 0.19521407079477998整体对比可视化如下所示：
整体测试识别准确率还不错，后续会再此基础上开发构建检测模型。