训练素材数据转变过程

数据转变过程(真实例子全解析)

本文用真实数据 一步步展示:samples/000.json 里的文本是怎么变成 dataset_hr.npz 里的二进制数组的。

所有数值都是实际跑出来的,不是编的。

一句话版本:

JSON 文本 → 字节 → numpy 图像数组 → 缩放归一化小数数组 → 和标签一起塞进 .npz(二进制压缩包)


0. 总览:一张图看懂数据流

复制代码
samples/000.json(文本文件,50KB)
        │  json.load()
        ▼
Python 字典 dict
        │  取 backgroundImage 字段(base64 字符串)
        ▼
"data:image/jpeg;base64,/9j/4AAQSkZJRgABA..."
        │  base64.b64decode()
        ▼
37622 个字节(bytes,JPEG 文件)
        │  cv2.imdecode()
        ▼
numpy 数组 (360, 600, 3)  uint8      ← 彩色图
        │  转灰度 → 缩放到 300x180 → 除以255
        ▼
numpy 数组 (180, 300)     float32    ← 背景小图
        │  np.stack([背景小图, 缺口mask小图])
        ▼
numpy 数组 (2, 180, 300)  float32    ← 1 个样本 X
        │  600 个样本 np.stack + 标签
        ▼
dataset_hr.npz  (X: (600,2,180,300), y: (600,), ids: (600,))

1. 原材料:真实 JSON 文件

文件 samples/000.json 的开头是这样的(截断显示):

json 复制代码
{"id": "SLIDER_a07b26a14a5b40848cb847b7eada8934", "type": "SLIDER", "backgroundImage": "data:image/jpeg;base64,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", "templateImage": "data:image/png;base64,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", "backgroundImageTag": "default", "templateImageTag": "default", "backgroundImageWidth": 600, "backgroundImageHeight": 360, "templateImageWidth": 110, "templateImageHeight": 360, "data": null}

关键点:这个文件本身只是文本。图片不是图,而是两串超长的 base64 字符。


2. 第 1 步:JSON 文本 → Python 字典

Python 用 json.load() 读文件,得到 dict(字典)

python 复制代码
import json

with open("samples/000.json", "r", encoding="utf-8") as f:
    data = json.load(f)

print(type(data))     # <class 'dict'>
print(list(data.keys()))
# ['id', 'type', 'backgroundImage', 'templateImage', 'backgroundImageTag',
#  'templateImageTag', 'backgroundImageWidth', 'backgroundImageHeight',
#  'templateImageWidth', 'templateImageHeight', 'data']
JSON 里的东西 变成 Python 里的东西
"SLIDER_a07b..." 字符串 str
600 整数 int
"data:image/jpeg;base64,..." 字符串 str(长度 50187 个字符

注意:这一步只是"文本变成内存对象",图片还是字符串,还没变成图。


3. 第 2 步:base64 字符串 → 原始字节(bytes)

backgroundImage 这个字段是 data:image/jpeg;base64, 前缀 + 一长串 base64 字符。

把它 split(",") 去掉前缀,再 base64.b64decode() 解码:

python 复制代码
bg_uri = data["backgroundImage"]          # 50187 个字符
raw = base64.b64decode(bg_uri.split(",")[1])

print(type(raw))       # <class 'bytes'>
print(len(raw))        # 37622  (字符 50187 -> 字节 37622,变短了)
print(list(raw[:16]))
# [255, 216, 255, 224, 0, 16, 74, 70, 73, 70, 0, 1, 2, 0, 0, 1]
阶段 类型 长度/大小 特征
JSON 里的字符串 str 50187 字符 可读文本,/9j/4AAQ...
解码后 bytes 37622 字节 前 2 字节是 255, 216(十六进制 FF D8

为什么 FF D8 很关键?因为 FF D8 FF 就是 JPEG 图片的文件头。这一步证明:字符串终于变回了"一张 jpg 文件的内容"。


4. 第 3 步:字节 → numpy 图像数组

用 OpenCV 的 cv2.imdecode() 把字节解码成 numpy 数组:

python 复制代码
import cv2

img = cv2.imdecode(np.frombuffer(raw, dtype=np.uint8), cv2.IMREAD_COLOR)

print(type(img))     # <class 'numpy.ndarray'>
print(img.dtype)     # uint8
print(img.shape)     # (360, 600, 3)   ← 高360、宽600、3个颜色通道(BGR)
print(img.min(), img.max())   # 0 ~ 255

此刻图才真正变成"数字"

复制代码
一张 600x360 的彩色图
= 360 行 × 600 列 × 3 个颜色值(B/G/R)
= 648000 个 0~255 之间的整数

在 numpy 里就是 (360, 600, 3) 的三维数组。

这样的数据

复制代码
[[[254 179 223]
  [254 179 223]
  [254 179 223]
  ...
  [155  44  48]
  [153  42  46]
  [151  40  44]]

 [[254 179 223]
  [254 179 223]
  [254 179 223]
  ...
  [156  45  49]
  [154  43  47]
  [152  41  45]]

 [[253 181 224]
  [253 181 224]
  [253 181 224]
  ...
  [156  45  49]
  [154  43  47]
  [153  42  46]]

 [[254 182 225]
  [254 182 225]
  [254 182 225]
  ...
  [156  45  49]
  [155  44  48]
  [154  43  47]]

 [[254 182 225]
  [254 182 225]
  [254 182 225]
  ...
  [156  45  49]
  [155  44  48]
  [154  43  47]]]

5. 第 4 步:图像 → 灰度 → 缩放 → 归一化小数

模型输入要求:300x180、双通道、0~1 的小数。所以继续处理:

python 复制代码
# 5.1 彩色 -> 灰度(去掉颜色,只剩明暗)
g = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)      # (360, 600) uint8
# 真实像素值(左上角 3x5):
#   [201, 201, 201, 201, 201]
#   [201, 201, 201, 201, 201]
#   [202, 202, 202, 202, 202]

# 5.2 缩放:600x360 -> 300x180
g = cv2.resize(g, (300, 180), interpolation=cv2.INTER_AREA)

# 5.3 除以 255:整数 0~255 -> 小数 0~1
bg_s = g.astype(np.float32) / 255.0

print(bg_s.dtype)    # float32
print(bg_s.shape)    # (180, 300)
# 真实像素值(同一个位置变成小数):
#   [0.7882, 0.7882, 0.7882, 0.7882, 0.7922]
#   [0.7961, 0.7961, 0.7961, 0.7961, 0.7961]
#   [0.8,   0.8,   0.8,   0.8,   0.8   ]
转换 例子
整数 201 201 / 255
小数 0.7882(float32)

为什么除以 255?神经网络的"旋钮(权重)"对 0~1 的小数学习得又快又稳,对 0~255 的大整数反应迟钝。


6. 第 5 步:模板图 → 缺口 mask(第二通道)

templateImage 是一张 PNG(带透明通道) 的缺口小图。模型不看它的颜色,只看"缺口在哪",所以取它的 alpha 透明通道 做成黑白 mask:

python 复制代码
tp = cv2.imdecode(..., cv2.IMREAD_UNCHANGED)   # (360, 110, 4)  4=BGRA
mask = (tp[:, :, 3] > 0).astype(np.uint8)      # 不透明=1,透明=0
m_s = cv2.resize(mask, (300, 180)).astype(np.float32) / 255.0

print(m_s.shape)    # (180, 300)  float32
print(m_s[:3, :5])  # 左上角全是 0.0(透明区域)
# [[0., 0., 0., 0., 0.],
#  [0., 0., 0., 0., 0.],
#  [0., 0., 0., 0., 0.]]

但注意真实代码prepare_dataset.py / rebuild_dataset.py 第 49 行)是:mask 先做成 0/1 二值,再除以了 255

python 复制代码
mask = (tp[:, :, 3] > 0).astype(np.uint8)                    # 0 或 1
m_s = cv2.resize(mask, (300, 180), ...).astype(np.float32) / 255.0   # 0 或 1/255

所以存进 npz 后的真实值不是 1.0,而是:

mask 位置 你可能以为 真实值(npz 实测)
缺口处 1.0 0.0039216(= 1/255)
其余 0.0 0.0

用 npz 实测:通道1 全局最大值 = 0.003921568859368563>0.003 的像素有 11709 个 (缺口轮廓),>0.5 的像素 0 个

为什么值这么小还管用?因为训练和预测用同一套代码 (都除以 255),缺口处 0.0039 vs 别处 0.0 的"相对差异"是稳定的,网络照样能学出缺口位置------这也解释了准确率为什么仍有 95%。想更规范(缺口=1.0)的话,把这几处 / 255.0 去掉重新训练即可,但不是必须的


7. 第 6 步:np.stack → 一个样本 X

把"背景小图"和"缺口mask小图"摞成一个 双通道 样本:

python 复制代码
x = np.stack([bg_s, m_s])

print(x.shape)    # (2, 180, 300)  float32
复制代码
通道0: 背景灰度图(180行 × 300列)
通道1: 缺口mask   (180行 × 300列)

1 张验证码 = 1 个这样的数组 。这是模型真正"看到"的最小单位,维度是 (通道, 高, 宽)


8. 第 7 步:标签 randomX → y

训练还需要"正确答案":缺口在 x 方向的像素位置。

服务端生成验证码时,缺口中心 x 坐标就是 randomX。例如真实数据里:

复制代码
服务端 randomX = 280   (600 尺度原图)

因为模型输入是 300 宽(正好是 600 的一半),标签也要跟着缩一半:

复制代码
y = randomX × (300 / 600) = 280 × 0.5 = 140.0

验证:dataset_hr.npzSLIDER_a07b26a14a5b40848cb847b7eada8934 这个样本的 y 正是 140.0

名称 说明
randomX 280 服务端真实缺口位置(600尺度)
y(存进 npz) 140.0 训练标签(300尺度)
换算公式 y = randomX / 2 图片缩一半,坐标跟着缩一半

推理/预测时正好反过来:模型输出的 300 尺度值 × 2 = 600 尺度像素位置,再除以背景图宽度比例换算成前端坐标。


9. 第 8 步:600 个样本 → np.savez → dataset_hr.npz

prepare_dataset.py 循环处理 samples/ 下的 600 个 json,每个都变成:

  • 一个 x(2, 180, 300) float32
  • 一个 y140.0 这样的 float32

然后全部拼接 + 打包:

python 复制代码
X   = np.stack(所有 x)      # (600, 2, 180, 300)  float32
y   = np.array(所有 y)      # (600,)              float32
ids = np.array(所有 id)     # (600,)              <U39(字符串)

np.savez("dataset_hr.npz", X=X, y=y, ids=ids)

真实结果:

类型 真实数值
X (600, 2, 180, 300) float32 600 张验证码的双通道图
y (600,) float32 600 个缺口坐标(300尺度),最小 116 / 最大 479(换算回600尺度)
ids (600,) 字符串 600 个验证码 id
复制代码
磁盘上 1 个文件:dataset_hr.npz ≈ 247 MB

10. npz 到底是什么?

不是 json,不是组合文本,是一个二进制压缩包(zip 格式)。

python 复制代码
np.load("dataset_hr.npz")   # 打开后就是一个 dict-like 对象
npz["X"].shape              # (600, 2, 180, 300)

内部就是"键 → 二进制数组"的映射,等价于一个 zip 里有 3 个文件:

复制代码
dataset_hr.npz(zip 容器)
├── X.npy    → 二进制数组 (600,2,180,300) float32
├── y.npy    → 二进制数组 (600,)          float32
└── ids.npy  → 二进制数组 (600,)          <U39
  • 二进制,不是文本,所以人类读不懂,numpy 一读就懂。
  • float32 比 float64 省一半空间,4 字节存一个数。
  • 存的是 0~1 的小数和标签坐标不再是 json 里的 base64 字符串。

11. 每一步对应的代码位置

步骤 做什么 代码位置
1 读 json → dict prepare_dataset.py / rebuild_dataset.py
2 base64 解码 slider_solver.py_decode_data_uri()
3 字节 → 图像数组 cv2.imdecode()
4 灰度/缩放/归一化 prepare_dataset.py
5 模板 → mask prepare_dataset.py
6 np.stack 双通道 prepare_dataset.py
7 randomX/2 → y prepare_dataset.py
8 600 样本 np.savez prepare_dataset.py

12. 自己动手验证(复制即用)

python 复制代码
import json, numpy as np, cv2

# 1) 读 json
data = json.load(open("samples/000.json", "r", encoding="utf-8"))
print("id:", data["id"])

# 2) base64 -> 字节
raw = data["backgroundImage"].split(",")[1].encode()
import base64
raw = base64.b64decode(raw)
print("解码后字节数:", len(raw), "| 文件头:", raw[:3].hex())

# 3) 字节 -> 图像
img = cv2.imdecode(np.frombuffer(raw, np.uint8), cv2.IMREAD_COLOR)
print("图像数组:", img.shape, img.dtype)

# 4) 归一化小图
bg_s = cv2.resize(cv2.cvtColor(img, cv2.COLOR_BGR2GRAY),
                  (300, 180)).astype(np.float32) / 255.0
print("归一化小图:", bg_s.shape, bg_s.dtype, "左上角:", bg_s[0, :5])

# 5) 从 npz 取出这个样本的 X(模型真正看到的输入)和标签
npz = np.load("dataset_hr.npz")
idx = list(npz["ids"]).index(data["id"])

x_sample = npz["X"][idx]                            # 这个样本的 X
print("该样本 X:", x_sample.shape, x_sample.dtype)  # (2, 180, 300) float32
print("通道0(背景) 左上角3x5:", x_sample[0, :3, :5])
print("通道1(mask) 左上角3x5:", x_sample[1, :3, :5])
print("通道1 全局最大值:", float(x_sample[1].max()))  # 0.0039,不是 1.0!
print("真实 y =", float(npz["y"][idx]), "-> 服务端 randomX =", float(npz["y"][idx]) * 2)
npz.close()

输出(真实):

复制代码
id: SLIDER_a07b26a14a5b40848cb847b7eada8934
解码后字节数: 37622 | 文件头: ffd8ff
图像数组: (360, 600, 3) uint8
归一化小图: (180, 300) float32 左上角: [0.7882 0.7882 0.7882 0.7882 0.7922]
该样本 X: (2, 180, 300) float32
通道0(背景) 左上角3x5: [[0.7882 0.7882 0.7882 0.7882 0.7922]
                        [0.7961 0.7961 0.7961 0.7961 0.7961]
                        [0.8    0.8    0.8    0.8    0.8   ]]
通道1(mask) 左上角3x5: [[0. 0. 0. 0. 0.]
                        [0. 0. 0. 0. 0.]
                        [0. 0. 0. 0. 0.]]
通道1 全局最大值: 0.003921568859368563
真实 y = 140.0 -> 服务端 randomX = 280.0

想看缺口位置(x≈140 列)的真实值,把上面 :3, :5 改成 60:80, 130:150 即可,

通道0 会在缺口凹槽处突然变暗(如 0.93 → 0.12),通道1 同时出现一片 0.0039


13. 终极一句话记忆

json 里的 base64 字符串 → 解成字节 → 解成像素数组 → 缩放归一化成 0~1 小数 → 和标签(randomX/2)一起用 np.savez 打包成二进制 .npz ------ 模型训练时读的就是这个 .npz。


14. 附录:一份样本的完整真实数据(从 JSON 到 NPZ 全程对照)

以下所有数据都是对同一个样本实际读取出来的,可直接和你的输出逐一比对。

样本:samples/000.json,id = SLIDER_a07b26a14a5b40848cb847b7eada8934

说明:一张图片有 54000~648000 个像素,不可能也没必要逐个贴出来;所以对图片 给出"完整统计 + 多个区域真实值",

标签 y 给出完整统计、分布和前 20 个值。想看任意区域,用第 12 节的脚本改下标即可。

14.1 原材料:JSON 的 11 个键(完整)

类型 长度 真实值
id str 39 SLIDER_a07b26a14a5b40848cb847b7eada8934
type str 6 SLIDER
backgroundImage str 50187 字符 开头 data:image/jpeg;base64,/9j/4AAQSkZJRgABAgAAAQABAAD...,结尾 ...UV6NCT5DN7n/9k=
templateImage str 25678 字符 开头 data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAG4A...,结尾 ...AAASUVORK5CYII=
backgroundImageTag str 7 default
templateImageTag str 7 default
backgroundImageWidth int - 600
backgroundImageHeight int - 360
templateImageWidth int - 110
templateImageHeight int - 360
data - - null(本样本无)

14.2 base64 解码:字符串 → 字节(完整)

复制代码
字符数:  50187
字节数:  37622
前16字节: [255, 216, 255, 224, 0, 16, 74, 70, 73, 70, 0, 1, 2, 0, 0, 1]
末尾16字节: [114, 14, 64, 254, 116, 81, 69, 122, 52, 36, 249, 12, 222, 231, 255, 217]
sha256(前16位): 6f5b6fc41321f1f6   ← 文件指纹,可用于核对两份数据是否同一张图

前2字节=255,216(十六进制 FF D8)= JPEG 文件头;末尾=255,217FF D9)= JPEG 文件尾。这是"完整的一张 jpg"。

14.3 背景图解码:字节 → numpy 数组(完整统计)

复制代码
shape: (360, 600, 3)   高360 × 宽600 × 3通道(BGR)
dtype: uint8           整数
像素总数: 648000
min: 0   |  max: 255   |  mean: 186.84

14.4 灰度 + 缩放 + 归一化(完整统计)

复制代码
灰度图   shape: (360, 600)   dtype: uint8  |  min: 2   |  max: 255
归一化   shape: (180, 300)   dtype: float32 |  像素总数: 54000
min: 0.0314  |  max: 1.0  |  mean: 0.7155  |  std: 0.2158

左上角真实值(float32 完整精度):

复制代码
[0.78820002, 0.78820002, 0.78820002, 0.78820002, 0.79220003]
[0.79610002, 0.79610002, 0.79610002, 0.79610002, 0.79610002]
[0.80000001, 0.80000001, 0.80000001, 0.80000001, 0.80000001]

14.5 模板 → 缺口 mask(完整统计)

复制代码
模板原图  shape: (360, 110, 4)   dtype: uint8   (4=BGRA,带透明通道)
mask(大)  唯一值: [0, 1]         非零像素: 8393 / 39600   ← 缺口轮廓占约 21%
mask(小)  shape: (180, 300)      最大值: 0.0039216
          >0.003 的像素: 11709 / 54000               ← 缩小后轮廓被"加宽"了

14.6 np.stack:合成 1 个样本 X(完整统计)

复制代码
shape: (2, 180, 300)   dtype: float32   总数值个数: 108000
min: 0.0  |  max: 1.0

缺口区域真实值(行 60~62,列 130~149)------注意通道0 在列 ~140 处突然变暗(缺口凹槽),通道1 同时出现 0.0039:

复制代码
通道0(背景):
[0.8314, 0.8353, 0.8353, 0.8353, 0.8353, 0.8353, 0.8392, 0.8353, 0.8353, 0.8353, 0.7882, 0.7059, 0.698, 0.6863, 0.651, 0.6667, 0.6588, 0.6549, 0.6392, 0.6627]
[0.9137, 0.9137, 0.9137, 0.9137, 0.9137, 0.9137, 0.9255, 0.9216, 0.9176, 0.9137, 0.8392, 0.7137, 0.6902, 0.6784, 0.6588, 0.6745, 0.6667, 0.6627, 0.6471, 0.6667]
[0.9216, 0.9255, 0.9216, 0.9216, 0.9216, 0.9216, 0.9216, 0.9216, 0.9176, 0.9176, 0.8549, 0.7137, 0.7098, 0.7059, 0.6667, 0.6824, 0.6745, 0.6706, 0.6549, 0.6706]

通道1(mask):
[0.0,    0.0,    0.0,    0.0,    0.0,    0.0,    0.0,    0.0,    0.0,    0.0,    0.0,    0.0,    0.0,    0.0,    0.0,    0.0,    0.0,    0.0,    0.0,    0.0]
[0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039]
[0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039, 0.0039]

14.7 dataset_hr.npz 完整结构

复制代码
键: ['X', 'y', 'ids']

X  : (600, 2, 180, 300)  float32   ← 600 个样本,总数值 64800000,占内存 247.2 MB
y  : (600,)              float32   ← 600 个标签(300尺度)
ids: (600,)              <U39      ← 600 个验证码 id(字符串,最长39字符)

npz 磁盘大小: 247.3 MB(比 nbytes 略大,因为还有头部/压缩开销)

14.8 该样本在 npz 里的 X 与 y

复制代码
X[idx]: (2, 180, 300) float32
y[idx]: 140.0   →  服务端 randomX = 280.0

14.9 y 标签全集统计(600 个,300尺度)

复制代码
min: 58.0   |  max: 239.5   |  mean: 151.5   |  std: 52.1
换算回600尺度: min 116.0 | max 479.0

前 20 个: [140.0, 134.5, 197.5, 89.0, 83.0, 60.0, 199.5, 115.0, 161.0, 155.5,
           59.0,  58.0,  95.5,  162.0, 213.5, 222.5, 168.0, 146.0, 99.5, 84.5]

分布(按600尺度分30桶,每桶约12px):
[28, 18, 14, 9, 13, 26, 15, 24, 22, 16, 23, 12, 27, 20, 16, 22, 21, 18,
 19, 18, 25, 24, 24, 21, 19, 24, 27, 18, 18, 19]
→ 各位置都有样本,分布比较均匀,说明 600 张数据覆盖了全图各个缺口位置

14.10 ids 全集

复制代码
类型: (600,) <U39    600 个全部唯一(True)

前 10 个:
['SLIDER_a07b26a14a5b40848cb847b7eada8934',
 'SLIDER_e1514bddb5ef417e9ea13c1c55818385',
 'SLIDER_1ac2ab435cb048b28f79bb155f099015',
 'SLIDER_0e6c1a48471342779996019272c1b68a',
 'SLIDER_03c23ad9bb7c4a66ae3ea09b722808eb',
 'SLIDER_e2e77e10791d41f09112f20663207f1e',
 'SLIDER_f7b00b242da9461bb6d041eaef191d9f',
 'SLIDER_393a45e87e754ba9aa37330d14b4a5b9',
 'SLIDER_c979eaead439466fb7b06b440bfb5ed3',
 'SLIDER_5352c60499f24f579b9253c203d31b44']

14.11 全量核对方法

想看任何区域、任何样本的完整真实数据,改第 12 节脚本的下标即可:

python 复制代码
# 看第 2 个样本 (idx=1) 的整张背景图右下角 5x5
x2 = np.load("dataset_hr.npz")["X"][1]
print(x2[0, -5:, -5:])      # 通道0 右下角
print(x2[1, -5:, -5:])      # 通道1 右下角
相关推荐
东莞市云毅网络有限公司14 分钟前
基于GEO的品牌信源矩阵构建:从内容生产到AI采信的全链路设计
人工智能·机器学习·矩阵
不吃鱼的羊14 分钟前
图式疗法简介
人工智能
测试199816 分钟前
UI自动化测试:窗口截图&文件上传实战
自动化测试·软件测试·python·selenium·测试工具·职场和发展·测试用例
叫我:松哥22 分钟前
基于Flask的教师评教管理系统,支持学生和教师两种角色登录
数据库·后端·python·数据挖掘·flask
成都佳洋光电科技有限公司23 分钟前
一文了解机器视觉成像
图像处理·人工智能·工业相机·工业镜头·短波红外
雷帝木木24 分钟前
开源与商业AI工具的选型框架:总拥有成本、定制化与社区活力评估
人工智能·python·深度学习·机器学习
IT古董25 分钟前
AI 资讯日报 | 2026年8月24日:资本加码全栈AI,开源模型持续反攻,价格战与算力涨价并行,行业竞争全面转向效率、生态与商业回报
人工智能
纽格立科技28 分钟前
语音搜不到台,问题不全在车厂
人工智能·车载系统·音视频·语音识别·信息与通信·传媒
2501_9127840830 分钟前
跨境建站避坑:为什么通用电商架构不适配反向代购业务
大数据·人工智能·架构·taoify