From a6b9531df51dc30003b08f63cbeadeaf8666a791 Mon Sep 17 00:00:00 2001 From: flt6 <1404262047@qq.com> Date: Fri, 16 May 2025 15:46:34 +0800 Subject: [PATCH] hsv_1 Former-commit-id: 57a8fb7799ea7543b4af1ea49626346d50546f82 --- .gitattributes | 3 + Auto_Ctrl/.gitattributes | 3 - Picture_Train/main.py | 72 ++++++++++++++++++++---- Picture_Train/model.pkl | 3 + Picture_Train/show.py | 29 ++++++++++ Picture_Train/train_2.py | 118 +++++++++++++++++++++++++++++++++++++++ 6 files changed, 215 insertions(+), 13 deletions(-) delete mode 100644 Auto_Ctrl/.gitattributes create mode 100644 Picture_Train/model.pkl create mode 100644 Picture_Train/show.py create mode 100644 Picture_Train/train_2.py diff --git a/.gitattributes b/.gitattributes index dfe0770..40304a7 100644 --- a/.gitattributes +++ b/.gitattributes @@ -1,2 +1,5 @@ # Auto detect text files and perform LF normalization * text=auto +*.exe filter=lfs diff=lfs merge=lfs -text +*.pth filter=lfs diff=lfs merge=lfs -text +*.pkl filter=lfs diff=lfs merge=lfs -text diff --git a/Auto_Ctrl/.gitattributes b/Auto_Ctrl/.gitattributes deleted file mode 100644 index 02329e6..0000000 --- a/Auto_Ctrl/.gitattributes +++ /dev/null @@ -1,3 +0,0 @@ -*.exe filter=lfs diff=lfs merge=lfs -text -*.pth filter=lfs diff=lfs merge=lfs -text -*.pkl filter=lfs diff=lfs merge=lfs -text diff --git a/Picture_Train/main.py b/Picture_Train/main.py index eb389a0..82f7da2 100644 --- a/Picture_Train/main.py +++ b/Picture_Train/main.py @@ -1,16 +1,68 @@ -# 这是一个示例 Python 脚本。 +import cv2 +import numpy as np +from pathlib import Path +from train_2 import train_model,prepare_data,predict +import joblib -# 按 Shift+F10 执行或将其替换为您的代码。 -# 按 双击 Shift 在所有地方搜索类、文件、工具窗口、操作和设置。 +inp = Path("data/train") +val = Path("data/val") +proc = Path("data/proc") +proc.mkdir(exist_ok=True) -def print_hi(name): - # 在下面的代码行中使用断点来调试脚本。 - print(f'Hi, {name}') # 按 Ctrl+F8 切换断点。 +def preproc(dir:Path): + d={} + for file in dir.glob("*/*.jpg"): + im = cv2.imread(file) + if im is None: + print(f"Error reading image: {file}") + continue + cl = file.parents[0].name + if cl not in d: + d[cl] = [] + hsv = cv2.cvtColor(im,cv2.COLOR_BGR2HSV) + mask = hsv[:,:,1] > 150 + # cor = np.argwhere(mask) + # y_min,x_min = cor.min(axis=0) + # y_max,x_max = cor.max(axis=0) + mask = mask[:,:,np.newaxis] + # cv2.findCoun + # im = im[y_min:y_max,x_min:x_max] + cnt = np.count_nonzero(mask) + hsv*=mask + h = round(np.sum(hsv[:,:,0])/cnt) + s = round(np.sum(hsv[:,:,1])/cnt) + v = round(np.sum(hsv[:,:,2])/cnt) + name = f"{h}_{s}_{v}.jpg" + d[cl].append((h,s,v)) + # (proc/cl).mkdir(exist_ok=True) + # cv2.imwrite(proc/cl/name,cv2.cvtColor(hsv,cv2.COLOR_HSV2BGR)) + return d +d:dict[str,list[tuple[int,int,int]]] = preproc(inp) +val:dict[str,list[tuple[int,int,int]]] = preproc(val) -# 按装订区域中的绿色按钮以运行脚本。 -if __name__ == '__main__': - print_hi('PyCharm') +print("数据预处理完成") -# 访问 https://www.jetbrains.com/help/pycharm/ 获取 PyCharm 帮助 +model, label_map = train_model(d) +print("训练完成") +joblib.dump(model, "model.pkl") + +# model = joblib.load("model.pkl") +X_train, y_train = prepare_data(d) +print(predict(model, label_map, d)) +print(predict(model, label_map, val)) +# print(f"\n训练集准确率: {model.score(X_train, y_train):.5f}") +# X_train, y_train = prepare_data(val) +# print(f"\n训练集准确率: {model.score(X_train, y_train):.5f}") +# model.predict() +from src_predict import predictor +suc = cnt = 0 +for file in Path("data/train").glob("*/*.jpg"): + cnt+=1 + pcl,_=predictor(file) + acl = file.parents[0].name + if acl == pcl: + suc+=1 + +print(f"预测准确率: {suc/cnt:.4f}") diff --git a/Picture_Train/model.pkl b/Picture_Train/model.pkl new file mode 100644 index 0000000..06a50a5 --- /dev/null +++ b/Picture_Train/model.pkl @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:409653310924dc977ea2ae5dd46042ef144f4c8500c460ba5ca5b0f5ce68bed8 +size 535673 diff --git a/Picture_Train/show.py b/Picture_Train/show.py new file mode 100644 index 0000000..94538aa --- /dev/null +++ b/Picture_Train/show.py @@ -0,0 +1,29 @@ + +import matplotlib.pyplot as plt +import random + + +# 创建一个 3D 图形 +fig = plt.figure() +ax = fig.add_subplot(111, projection='3d') + +# 为每个 key 分配一个随机颜色 +colors = {"orange":(1,0,0),"yellow":(0,1,0)} +# for key in d.keys(): +# colors[key] = (random.random(), random.random(), random.random()) # 随机 RGB 颜色 + +# 绘制每个 key 的点 +for key, points in d.items(): + x_vals = [point[0] for point in points] + y_vals = [point[1] for point in points] + z_vals = [point[2] for point in points] + ax.scatter(x_vals, y_vals, z_vals, label=key, color=colors[key]) + +# 添加图例和标签 +ax.set_xlabel('X轴') +ax.set_ylabel('Y轴') +ax.set_zlabel('Z轴') +ax.legend() + +# 显示图形 +plt.show() diff --git a/Picture_Train/train_2.py b/Picture_Train/train_2.py new file mode 100644 index 0000000..8b841ed --- /dev/null +++ b/Picture_Train/train_2.py @@ -0,0 +1,118 @@ + +import numpy as np +from sklearn.ensemble import RandomForestClassifier +from typing import Dict, List, Tuple +import joblib + +def prepare_data(data: Dict[str, List[np.ndarray]]) -> Tuple[np.ndarray, np.ndarray]: + """ + 将dict[str, list[ndarray]]格式的数据转换为模型可用的特征矩阵和标签向量 + + 参数: + data: 格式为 dict[str, list[ndarray]] 的数据，其中键为类别名，值为对应类别的特征数组列表 + + 返回: + X: 特征矩阵 + y: 标签向量 + """ + features = [] + labels = [] + + # 为每个类别分配一个数字标签 + label_map = {class_name: i for i, class_name in enumerate(data.keys())} + + for class_name, arrays_list in data.items(): + label = label_map[class_name] + for arr in arrays_list: + # 处理每个数组中的每个样本 + features.append(np.array(arr)) + labels.append(label) + # if len(arr.shape) > 1: + # for sample in arr: + # features.append(sample) + # labels.append(label) + # else: + # # 处理单个样本的情况 + # features.append(arr) + # labels.append(label) + + return np.array(features), np.array(labels) + +def train_model(data: Dict[str, List[np.ndarray]]): + """ + 训练分类模型 + + 参数: + data: 训练数据，格式为 dict[str, list[ndarray]] + + 返回: + 训练好的模型和标签映射字典 + """ + X, y = prepare_data(data) + + # 创建并训练模型 + model = RandomForestClassifier(n_estimators=100, random_state=42) + model.fit(X, y) + + # 创建逆向映射，用于将数字标签转回类别名 + label_map = {i: class_name for i, class_name in enumerate(data.keys())} + + return model, label_map + +def predict(model, label_map: Dict[int, str], val_data: Dict[str, List[np.ndarray]]) -> Dict[str, List[List[str]]]: + """ + 使用训练好的模型对验证数据进行预测 + + 参数: + model: 训练好的模型 + label_map: 标签映射字典，用于将数字标签转换回类别名 + val_data: 验证数据，格式为 dict[str, list[ndarray]] + + 返回: + 预测结果字典，格式为 dict[str, list[list[str]]]，表示每个输入数组中样本的预测类别 + """ + results = {} + + suc = 0 + cnt = 0 + for class_name, arrays_list in val_data.items(): + class_predictions = [] + for arr in arrays_list: + # 确保数据格式正确 + arr = np.array(arr) + cnt+=1 + if len(arr.shape) == 1: + arr = arr.reshape(1, -1) + + # 进行预测并转换为类别名 + pred_labels = model.predict(arr) + pred_classes = [label_map[label] for label in pred_labels] + if len(pred_classes) > 1:continue + if class_name==pred_classes[0]: + suc+=1 + # class_predictions.append(pred_classes) + + results[class_name] = class_predictions + + return suc/cnt + +if __name__ == "__main__": + exit() + # 训练模型 + model, label_map = train_model(d) + print("训练完成") + joblib.dump(model, "model.pkl") + + # 在验证数据上进行预测 + # predictions = predict(model, label_map, val) + + # 输出预测结果 + # print("预测结果:") + # for class_name, class_preds in predictions.items(): + # print(f"{class_name}:") + # for i, arr_preds in enumerate(class_preds): + # print(f" 数组 {i}: {arr_preds}") + + # 输出模型性能评估 + # X_train, y_train = prepare_data(val) + # print(f"\n训练集准确率: {model.score(X_train, y_train):.4f}") \ No newline at end of file