swoopyy

virtualenv venv -p python3.7
source venv/bin/activate
pip3 install torch torchvision
pip3 install opencv-python
pip3 install 'git+https://github.com/facebookresearch/fvcore'
pip3 install cython; pip install 'git+https://github.com/cocodataset/cocoapi.git#subdirectory=PythonAPI'

swoopyy

from detectron2.utils.visualizer import ColorMode
dataset_dicts = get_balloon_dicts("balloon/val")
for d in random.sample(dataset_dicts, 3):    
    im = cv2.imread(d["file_name"])
    outputs = predictor(im)
    v = Visualizer(im[:, :, ::-1],
                   metadata=balloon_metadata, 
                   scale=0.8, 
                   instance_mode=ColorMode.IMAGE_BW   # делаем фон черно-белым
    )

swoopyy

cfg.MODEL.WEIGHTS = os.path.join(cfg.OUTPUT_DIR, "model_final.pth")
cfg.MODEL.ROI_HEADS.SCORE_THRESH_TEST = 0.5 
cfg.DATASETS.TEST = ("balloon/val", )
predictor = DefaultPredictor(cfg)

swoopyy

from detectron2.engine import DefaultTrainer
from detectron2.config import get_cfg

cfg = get_cfg()
cfg.merge_from_file("./detectron2_repo/configs/COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x.yaml")
cfg.DATASETS.TRAIN = ("balloon/train",)
cfg.DATASETS.TEST = ()   # на данном этапе производить оценку качества не будем
cfg.DATALOADER.NUM_WORKERS = 2
cfg.MODEL.WEIGHTS = "detectron2://COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x/137849600/model_final_f10217.pkl"  
cfg.SOLVER.IMS_PER_BATCH = 2

swoopyy

import random

dataset_dicts = get_balloon_dicts("balloon/train")
for d in random.sample(dataset_dicts, 3):
    img = cv2.imread(d["file_name"])
    visualizer = Visualizer(img[:, :, ::-1], metadata=balloon_metadata, scale=0.5)
    vis = visualizer.draw_dataset_dict(d)
    cv2_imshow(vis.get_image()[:, :, ::-1])

swoopyy

import os
import numpy as np
import json
from detectron2.structures import BoxMode
import itertools

# функция загрузки изображений в стандартный для Detectron2 формат
def get_balloon_dicts(img_dir):
    # COCO json аннотации
    json_file = os.path.join(img_dir, "via_region_data.json")

swoopyy

# скачивание и распаковывание датасета
!wget https://github.com/matterport/Mask_RCNN/releases/download/v2.1/balloon_dataset.zip
!unzip balloon_dataset.zip > /dev/null

swoopyy

KEYPOINT_CONFIG = "./detectron2_repo/configs/COCO-Keypoints/keypoint_rcnn_R_101_FPN_3x.yaml"
KEYPOINT_CONFIG_WEIGHTS = "detectron2://COCO-Keypoints/keypoint_rcnn_R_101_FPN_3x/138363331/model_final_997cc7.pkl"

predictor = get_predictor(KEYPOINT_CONFIG, KEYPOINT_CONFIG_WEIGHTS)
outputs = predictor(im)
visualise(outputs)

swoopyy

DEFAULT_CONFIG = "./detectron2_repo/configs/COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x.yaml"
DEFAULT_CONFIG_WEIGHTS = "detectron2://COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x/137849600/model_final_f10217.pkl"

cfg = get_cfg()

# Метод конфигурации модели с заданными параметрами и весами
def get_predictor(config_path=DEFAULT_CONFIG, config_weights=DEFAULT_CONFIG_WEIGHTS):
  cfg.merge_from_file(config_path)
  cfg.MODEL.ROI_HEADS.SCORE_THRESH_TEST = 0.5  # Порог для заданной модели
  cfg.MODEL.WEIGHTS = config_weights

swoopyy

predictor = get_predictor()
outputs = predictor(im)
visualise(outputs)