Tensorflow对象检测API-错误的对象检测

Question

在Google-colab中，我尝试使用带有SSD_mobilenet_v1_pets.config的 Tensorflow对象检测API 来检测汽车，它将humans检测为car和{{1 }}设为car。以下是N/A和size config：

image dimensions

我有1160张各种尺寸的图像（例如：73 x 63、118 x 62、62 x 56、71 x 56、276 x 183、259 x 184、318 x 159、700 x 420、647 x 407、897 x 554）

我上面提到的输出：

请澄清，错误检测汽车的原因是由于图像尺寸还是其他原因？

这是我的配置文件

anchor_generator {
  ssd_anchor_generator {
    num_layers: 6
    min_scale: 0.2
    max_scale: 0.95
    aspect_ratios: 1.0
    aspect_ratios: 2.0
    aspect_ratios: 0.5
    aspect_ratios: 3.0
    aspect_ratios: 0.3333
  }
}
image_resizer {
  fixed_shape_resizer {
    height: 300
    width: 300
  }

这是我的pbtxt代码

    model {
  ssd {
    num_classes: 1
    box_coder {
      faster_rcnn_box_coder {
        y_scale: 10.0
        x_scale: 10.0
        height_scale: 5.0
        width_scale: 5.0
      }
    }
    matcher {
      argmax_matcher {
        matched_threshold: 0.5
        unmatched_threshold: 0.5
        ignore_thresholds: false
        negatives_lower_than_unmatched: true
        force_match_for_each_row: true
      }
    }
    similarity_calculator {
      iou_similarity {
      }
    }
    anchor_generator {
      ssd_anchor_generator {
        num_layers: 6
        min_scale: 0.2
        max_scale: 0.95
        aspect_ratios: 1.0
        aspect_ratios: 2.0
        aspect_ratios: 0.5
        aspect_ratios: 3.0
        aspect_ratios: 0.3333
      }
    }
    image_resizer {
      fixed_shape_resizer {
        height: 300
        width: 300
      }
    }
    box_predictor {
      convolutional_box_predictor {
        min_depth: 0
        max_depth: 0
        num_layers_before_predictor: 0
        use_dropout: false
        dropout_keep_probability: 0.8
        kernel_size: 1
        box_code_size: 4
        apply_sigmoid_to_scores: false
        conv_hyperparams {
          activation: RELU_6,
          regularizer {
            l2_regularizer {
              weight: 0.00004
            }
          }
          initializer {
            truncated_normal_initializer {
              stddev: 0.03
              mean: 0.0
            }
          }
          batch_norm {
            train: true,
            scale: true,
            center: true,
            decay: 0.9997,
            epsilon: 0.001,
          }
        }
      }
    }
    feature_extractor {
      type: 'ssd_mobilenet_v1'
      min_depth: 16
      depth_multiplier: 1.0
      conv_hyperparams {
        activation: RELU_6,
        regularizer {
          l2_regularizer {
            weight: 0.00004
          }
        }
        initializer {
          truncated_normal_initializer {
            stddev: 0.03
            mean: 0.0
          }
        }
        batch_norm {
          train: true,
          scale: true,
          center: true,
          decay: 0.9997,
          epsilon: 0.001,
        }
      }
    }
    loss {
      classification_loss {
        weighted_sigmoid {
          anchorwise_output: true
        }
      }
      localization_loss {
        weighted_smooth_l1 {
          anchorwise_output: true
        }
      }
      hard_example_miner {
        num_hard_examples: 3000
        iou_threshold: 0.99
        loss_type: CLASSIFICATION
        max_negatives_per_positive: 3
        min_negatives_per_image: 0
      }
      classification_weight: 1.0
      localization_weight: 1.0
    }
    normalize_loss_by_num_matches: true
    post_processing {
      batch_non_max_suppression {
        score_threshold: 1e-8
        iou_threshold: 0.6
        max_detections_per_class: 100
        max_total_detections: 100
      }
      score_converter: SIGMOID
    }
  }
}

train_config: {
  batch_size: 32
  optimizer {
    rms_prop_optimizer: {
      learning_rate: {
        exponential_decay_learning_rate {
          initial_learning_rate: 0.004
          decay_steps: 800720
          decay_factor: 0.95
        }
      }
      momentum_optimizer_value: 0.9
      decay: 0.9
      epsilon: 1.0
    }
  }
  fine_tune_checkpoint: "ssd_mobilenet_v1_coco_11_06_2017/model.ckpt"
  from_detection_checkpoint: true
  data_augmentation_options {
    random_horizontal_flip {
    }
  }
  data_augmentation_options {
    ssd_random_crop {
    }
  }
}

train_input_reader: {
  tf_record_input_reader {
    input_path: "object_detection/data/train.record"
  }
  label_map_path: "object_detection/data/object-detection.pbtxt"
}

eval_config: {
  num_examples: 40
}

eval_input_reader: {
  tf_record_input_reader {
    input_path: "object_detection/data/test.record"
  }
  label_map_path: "training/object-detection.pbtxt"
  shuffle: false
  num_readers: 1
}

我还有一个疑问，你们能帮我吗     我试图找出不戴头盔的人。我使用上述相同的模型。 这是我的pbtxt文件

  item {
  id: 1
  name: 'car'
}

我得到了这个输出

请帮助我...

Answer 1

请发布您的标签地图<?php /** * The file is controller. Do not modify the file if you want to upgrade the module in future * * @author Globo Jsc <contact@globosoftware.net> * @copyright 2016 Globo., Jsc * @link http://www.globosoftware.net * @license please read license in file license.txt */ include_once(_PS_MODULE_DIR_ . 'cardelivery/classes/AdditionalServicesModel.php'); class AdminAdditionalServiceController extends ModuleAdminControllerCore { public $name; public function __construct() { $this->name = 'AdminAdditionalService'; $this->className = 'AdditionalServicesModel'; $this->table = 'additional_service'; $this->meta_title = $this->l('Additional Services'); $this->deleted = false; $this->explicitSelect = true; $this->context = Context::getContext(); $this->bootstrap = true; $this->_defaultOrderBy = 'id_additional_service'; $this->filter = true; if (Shop::isFeatureActive()) { Shop::addTableAssociation($this->table, array('type' => 'shop')); } $this->position_identifier = 'id_additional_service'; $this->addRowAction('edit'); $this->addRowAction('delete'); $this->fields_list = array( 'id_additional_service' => array( 'title' => $this->l('ID'), 'type' => 'int', 'width' => 'auto', 'orderby' => false), 'service_name' => array( 'title' => $this->l('Icon'), 'width' => 'auto', 'orderby' => false, ), 'service_desc' => array( 'title' => $this->l('service_desc'), 'type' => 'text' ), 'active' => array( 'title' => $this->l('Status'), 'width' => 'auto', 'active' => 'status', 'type' => 'bool', 'orderby' => false), ); parent::__construct(); } function initContent() { parent::initContent(); if (Tools::isSubmit('submit')) { Tools::redirectAdmin(self::$currentIndex . '&token=' . Tools::getAdminTokenLite('AdminCategories') . '&conf=7'); } } public function initPageHeaderToolbar() { $this->page_header_toolbar_btn['back_to_list'] = array( 'href' => Context::getContext()->link->getAdminLink('AdminGCardeliverycity', true), 'desc' => $this->l('Back to list', null, null, false), 'icon' => 'process-icon-back' ); parent::initPageHeaderToolbar(); } public function renderForm() { $id_citydelivery = (int) Tools::getValue('id_citydelivery'); if ($id_citydelivery == 0) { $addSerModObj = new AdditionalServicesModel((int) Tools::getValue('id_additional_service')); $id_citydelivery = $addSerModObj->id_citydelivery; } $fields_form_1 = array( 'form' => array( 'legend' => array('title' => $this->l('Additional Service'), 'icon' => 'icon-cogs'), 'input' => array( array( 'type' => 'hidden', 'name' => 'id_citydelivery' ), array( 'type' => 'text', 'label' => $this->l('Service_name'), 'name' => 'service_name', 'size' => 255, 'required' => true, 'desc' => $this->l('Enter name of Arrival port') ), array( 'type' => 'text', 'label' => $this->l('service_desc'), 'name' => 'service_desc', 'size' => 255, 'required' => true, 'desc' => $this->l('Enter name of Arrival port') ), array( 'type' => 'text', 'label' => $this->l('charge'), 'name' => 'charge', 'size' => 255, 'required' => true, 'desc' => $this->l('Enter name of Arrival port') ), array( 'type' => 'switch', 'label' => $this->l('Active'), 'name' => 'active', 'required' => false, 'is_bool' => true, 'values' => array(array( 'id' => 'active_on', 'value' => 1, 'label' => $this->l('Active')), array( 'id' => 'active_off', 'value' => 0, 'label' => $this->l('Inactive')))), ), 'submit' => array('title' => $this->l('Save')), 'buttons' => array( array( 'href' => Context::getContext()->link->getAdminLink('AdminGCardeliverycity', true) . '&updatecitydelivery&id_citydelivery=' . $id_citydelivery, 'title' => $this->l('Cancle'), 'icon' => 'process-icon-cancel' ) ) ) ); $helper = new HelperForm(); $helper->show_toolbar = false; $helper->module = $this; $helper->name_controller = $this->name; $helper->toolbar_scroll = true; $lang = new Language((int) Configuration::get('PS_LANG_DEFAULT')); $helper->default_form_language = $lang->id; $helper->allow_employee_form_lang = Configuration::get('PS_BO_ALLOW_EMPLOYEE_FORM_LANG') ? Configuration::get('PS_BO_ALLOW_EMPLOYEE_FORM_LANG') : 0; $this->fields_form = array(); $helper->identifier = $this->identifier; $helper->submit_action = 'submit'; $helper->currentIndex = AdminController::$currentIndex; $helper->token = Tools::getAdminTokenLite($this->name); $id_additional_service = (int) Tools::getValue('id_additional_service'); $additionalServiceObj = new AdditionalServicesModel($id_additional_service); $helper->tpl_vars = array( 'fields_value' => $this->getFormValues($additionalServiceObj), 'languages' => $this->context->controller->getLanguages(), 'id_language' => $this->context->language->id ); $_1 = $helper->generateForm(array($fields_form_1)); $return = $_1; return $return; } function getFormValues($additionalServiceObj) { return array( 'service_name' => Tools::getValue('service_name ', $additionalServiceObj->service_name), 'service_desc' => Tools::getValue('service_desc', $additionalServiceObj->service_desc), 'charge' => Tools::getValue('charge', $additionalServiceObj->charge), 'active' => Tools::getValue('active', $additionalServiceObj->active) ); } } 。

我猜第一位只提到了赛车！！

Answer 2

正如@Janikan指出的那样，问题出在.pbtxt文件上。由于您使用的是默认的ssd_mobilenet模型，因此在MS-COCO数据集上进行了训练，该数据集实际上有90个类别，汽车的ID为3。由于在标签映射中找不到ID 3，因此输出显示为N /一种。默认标签图中的ID 1是person，这就是为什么其将“ car”显示为所有人的分类的原因。

如果您只想显示汽车。替换pbtxt文件并编辑visualisation_tools以仅过滤所需的class_Id。