chore: 战斗识别，删漏了的文件

src/MaaCore/Vision/Battle/BattleOperatorsImageAnalyzer.cpp

@@ -1,121 +0,0 @@
-#include "BattleOperatorsImageAnalyzer.h"
-
-#include <algorithm>
-#include <array>
-#include <cmath>
-
-#include "Utils/NoWarningCV.h"
-#include <onnxruntime/core/session/onnxruntime_cxx_api.h>
-
-#include "Config/Miscellaneous/TilePack.h"
-#include "Config/OnnxSessions.h"
-#include "Config/TaskData.h"
-#include "Utils/Logger.hpp"
-
-bool asst::BattleOperatorsImageAnalyzer::analyze()
-{
-    LogTraceFunction;
-
-
-    const double x_scale = 640.0 / m_image.cols;
-    const double y_scale = 640.0 / m_image.rows;
-
-    cv::Mat image;
-    cv::resize(m_image, image, cv::Size(), x_scale, y_scale, cv::INTER_AREA);
-    std::vector<float> input = image_to_tensor(image);
-
-    auto memory_info = Ort::MemoryInfo::CreateCpu(OrtDeviceAllocator, OrtMemTypeCPU);
-    constexpr int64_t batch_size = 1;
-    std::array<int64_t, 4> input_shape { batch_size, image.channels(), image.cols, image.rows };
-
-    Ort::Value input_tensor = Ort::Value::CreateTensor<float>(memory_info, input.data(), input.size(),
-                                                              input_shape.data(), input_shape.size());
-
-    auto& session = OnnxSessions::get_instance().get("operators_det");
-
-    Ort::AllocatorWithDefaultOptions allocator;
-    std::string input_name = session.GetInputNameAllocated(0, allocator).get();
-    std::string output_name = session.GetOutputNameAllocated(0, allocator).get();
-    std::vector input_names = { input_name.c_str() };
-    std::vector output_names = { output_name.c_str() };
-
-    Ort::RunOptions run_options;
-    auto output_tensors = session.Run(run_options, input_names.data(), &input_tensor, input_names.size(),
-                                      output_names.data(), output_names.size());
-
-    const float* raw_output = output_tensors[0].GetTensorData<float>();
-    // output_shape is { 1, 5, 8400 }
-    std::vector<int64_t> output_shape = output_tensors[0].GetTensorTypeAndShapeInfo().GetShape();
-
-    // yolov8 的 onnx 输出和前面的 v5, v7 等似乎不太一样，目前网上 yolov8 的 demo 较少，文档也没找到
-    // 这里的输出解析是我跟着数据推测的：
-    // center_x0, center_x1, ..... center_x8399
-    // center_y0, center_y1, ..... center_y8399
-    // w0, w1, ..... w8399
-    // h0, h1, ..... h8399
-    // conf0, conf1, ..... conf8399
-    // 如果后面要做多分类，可能得再看下怎么改（我也不知道shape会变成啥样）
-    std::vector<std::vector<float>> output(output_shape[1]);
-    for (int64_t i = 0; i < output_shape[1]; i++) {
-        output[i] = std::vector<float>(raw_output + i * output_shape[2], raw_output + (i + 1) * output_shape[2]);
-    }
-
-    std::vector<Box> all_results;
-    const auto& conf_vec = output.back();
-    for (size_t i = 0; i < conf_vec.size(); ++i) {
-        float score = conf_vec[i];
-        constexpr float Threshold = 0.3f;
-        if (score < Threshold) {
-            continue;
-        }
-        int center_x = static_cast<int>(output[0][i] / x_scale);
-        int center_y = static_cast<int>(output[1][i] / y_scale);
-        int w = static_cast<int>(output[2][i] / x_scale);
-        int h = static_cast<int>(output[3][i] / y_scale);
-
-        int x = center_x - w / 2;
-        int y = center_y - h / 2;
-        Rect rect { x, y, w, h };
-        all_results.emplace_back(Box { Cls::Operator, rect, score });
-    }
-
-    // NMS
-    constexpr double NmsThreshold = 0.7f;
-    std::sort(all_results.begin(), all_results.end(), [](const Box& a, const Box& b) { return a.score > b.score; });
-    std::vector<Box> nms_results;
-    for (size_t i = 0; i < all_results.size(); ++i) {
-        const auto& box = all_results[i];
-        if (box.score < 0.1f) {
-            continue;
-        }
-        nms_results.emplace_back(box);
-        for (size_t j = i + 1; j < all_results.size(); ++j) {
-            auto& box2 = all_results[j];
-            if (box2.score < 0.1f) {
-                continue;
-            }
-            int iou_area = (make_rect<cv::Rect>(box.rect) & make_rect<cv::Rect>(box2.rect)).area();
-            if (iou_area > NmsThreshold * box2.rect.area()) {
-                box2.score = 0;
-            }
-        }
-    }
-
-#ifdef ASST_DEBUG
-    int draw_offset_y = static_cast<int>(m_image.rows * -0.15);
-    int draw_offset_h = static_cast<int>(m_image.rows * 0.13);
-    for (const auto& box : nms_results) {
-        Rect draw_rect = box.rect;
-        draw_rect.y += draw_offset_y;
-        draw_rect.height += draw_offset_h;
-        cv::rectangle(m_image_draw, make_rect<cv::Rect>(draw_rect), cv::Scalar(0, 0, 255), 5);
-        cv::putText(m_image_draw, std::to_string(box.score), cv::Point(draw_rect.x, draw_rect.y - 10),
-                    cv::FONT_HERSHEY_PLAIN, 1.2, cv::Scalar(0, 0, 255), 2);
-        m_draw_rect.emplace_back(draw_rect);
-    }
-#endif
-
-    m_results = std::move(nms_results);
-
-    return true;
-}