pytorch分类模型绘制混淆矩阵以及可视化详解

#首先定义一个 分类数*分类数 的空混淆矩阵
 conf_matrix = torch.zeros(Emotion_kinds, Emotion_kinds)
 # 使用torch.no_grad()可以显著降低测试用例的GPU占用
    with torch.no_grad():
        for step, (imgs, targets) in enumerate(test_loader):
            # imgs:     torch.Size([50, 3, 200, 200])   torch.FloatTensor
            # targets:  torch.Size([50, 1]),     torch.LongTensor  多了一维，所以我们要把其去掉
            targets = targets.squeeze()  # [50,1] ----->  [50]

            # 将变量转为gpu
            targets = targets.cuda()
            imgs = imgs.cuda()
            # print(step,imgs.shape,imgs.type(),targets.shape,targets.type())
            
            out = model(imgs)
            #记录混淆矩阵参数
            conf_matrix = confusion_matrix(out, targets, conf_matrix)
            conf_matrix=conf_matrix.cpu()

def confusion_matrix(preds, labels, conf_matrix):
    preds = torch.argmax(preds, 1)
    for p, t in zip(preds, labels):
        conf_matrix[p, t] += 1
    return conf_matrix

conf_matrix=np.array(conf_matrix.cpu())# 将混淆矩阵从gpu转到cpu再转到np
corrects=conf_matrix.diagonal(offset=0)#抽取对角线的每种分类的识别正确个数
per_kinds=conf_matrix.sum(axis=1)#抽取每个分类数据总的测试条数

 print("混淆矩阵总元素个数：{0},测试集总个数:{1}".format(int(np.sum(conf_matrix)),test_num))
 print(conf_matrix)

 # 获取每种Emotion的识别准确率
 print("每种情感总个数：",per_kinds)
 print("每种情感预测正确的个数：",corrects)
 print("每种情感的识别准确率为：{0}".format([rate*100 for rate in corrects/per_kinds]))

# 绘制混淆矩阵
Emotion=8#这个数值是具体的分类数，大家可以自行修改
labels = ['neutral', 'calm', 'happy', 'sad', 'angry', 'fearful', 'disgust', 'surprised']#每种类别的标签

# 显示数据
plt.imshow(conf_matrix, cmap=plt.cm.Blues)

# 在图中标注数量/概率信息
thresh = conf_matrix.max() / 2	#数值颜色阈值，如果数值超过这个，就颜色加深。
for x in range(Emotion_kinds):
    for y in range(Emotion_kinds):
        # 注意这里的matrix[y, x]不是matrix[x, y]
        info = int(conf_matrix[y, x])
        plt.text(x, y, info,
                 verticalalignment='center',
                 horizontalalignment='center',
                 color="white" if info > thresh else "black")
                 
plt.tight_layout()#保证图不重叠
plt.yticks(range(Emotion_kinds), labels)
plt.xticks(range(Emotion_kinds), labels,rotation=45)#X轴字体倾斜45°
plt.show()
plt.close()