def run(
path: str,
key: str = 'measure',
show_image: bool = False,
show_video: bool = True,
):
dataset = COCO_Dataset.load_from_path(json_path=f'{path}/{key}-coco.json',
img_dir=path,
strict=False)
# for i, coco_ann in enumerate(dataset.annotations):
# if i % 2 == 0:
# coco_ann.segmentation = Segmentation()
# if i % 3 == 0:
# coco_ann.keypoints = Keypoint2D_List()
# coco_ann.num_keypoints = 0
for coco_ann in dataset.annotations:
coco_ann.segmentation = Segmentation()
coco_ann.keypoints = Keypoint2D_List()
coco_ann.num_keypoints = 0
coco_ann.keypoints_3d = None
coco_ann.camera = None
for coco_cat in dataset.categories:
coco_cat.keypoints = []
coco_cat.skeleton = []
dataset.save_to_path('non_strict_dataset.json',
overwrite=True,
strict=False)
dataset0 = COCO_Dataset.load_from_path('non_strict_dataset.json',
strict=False)
dataset0.images.sort(attr_name='file_name')
if show_image:
dataset0.save_visualization(save_dir=f'{path}_{dt_string3}_visualize',
show_preview=True,
kpt_idx_offset=-1,
overwrite=True,
show_details=True)
if show_video:
dataset0.save_video(
save_path=f'{path}_{dt_string3}.mp4',
# draw_order=['screw'],
show_details=True,
show_preview=True,
kpt_idx_offset=-1,
overwrite=True,
fps=5,
show_seg=True)
logger.green('Visualisation complete')
def combine_dataset_from_config_file(config_path: str, dest_folder_img: str,
dest_json_file: str):
# combine dataset
combined_dataset = COCO_Dataset.combine_from_config(
config_path=config_path,
img_sort_attr_name='file_name',
show_pbar=True)
combined_dataset.move_images(dst_img_dir=dest_folder_img,
preserve_filenames=False,
update_img_paths=True,
overwrite=True,
show_pbar=True)
combined_dataset.save_to_path(save_path=dest_json_file, overwrite=True)
def coco_json_fixer(coco_img_dir: str = None, fix_status: bool = True) -> str:
# def coco_json_fixer(coco_img_dir:str = None, blend_coco_anno:str = None, new_coco_anno:str = None, fix_status:bool=True):
if fix_status:
#fix coco
org_coco_file = os.path.join(coco_img_dir, "coco_annotations.json")
chk_file_exists(org_coco_file)
fixed_coco_file = os.path.join(coco_img_dir,
"fixed_coco_annotations.json")
dataset = COCO_Dataset.load_from_path(org_coco_file,
strict=False,
img_dir=coco_img_dir)
dataset.save_to_path(fixed_coco_file, strict=True, overwrite=True)
seg_filter(fixed_coco_file)
return fixed_coco_file
from annotation_utils.coco.structs import COCO_Dataset
dataset = COCO_Dataset.load_from_path(
json_path=
'/home/clayton/workspace/prj/data_keep/data/sekisui/hook/coco/output.json',
img_dir='/home/clayton/workspace/prj/data_keep/data/sekisui/hook/img')
for coco_image in dataset.images:
anns = dataset.annotations.get_annotations_from_imgIds([coco_image.id])
for ann in anns:
ann.bbox = ann.bbox.scale_about_center(
scale_factor=1.3,
frame_shape=[coco_image.height, coco_image.width])
dataset.save_to_path('bbox_resized.json', overwrite=True)
dataset.display_preview(show_details=True, kpt_idx_offset=-1)
from logger import logger
from annotation_utils.coco.structs import COCO_Dataset
dataset = COCO_Dataset.load_from_path(
json_path='bk_28_02_2020_11_18_30_coco-data/HSR-coco.json',
img_dir='bk_28_02_2020_11_18_30_coco-data')
logger.purple(
f'Flag0 len(dataset.images): {len(dataset.images)}, len(dataset.annotations): {len(dataset.annotations)}'
)
dataset.prune_keypoints(min_num_kpts=11, verbose=True)
logger.purple(
f'Flag1 len(dataset.images): {len(dataset.images)}, len(dataset.annotations): {len(dataset.annotations)}'
)
dataset.move_images(dst_img_dir='test_img',
preserve_filenames=True,
update_img_paths=True,
overwrite=True,
show_pbar=True)
logger.purple(
f'Flag2 len(dataset.images): {len(dataset.images)}, len(dataset.annotations): {len(dataset.annotations)}'
)
dataset.save_to_path(save_path='prune_test.json', overwrite=True)
dataset.display_preview(kpt_idx_offset=-1)
from common_utils.common_types.keypoint import Keypoint2D_List, Keypoint2D
# import printj
from imageaug import AugHandler, Augmenter as aug
from random import choice
from tqdm import tqdm
# import imgaug.augmenters as iaa
# PATH='/home/jitesh/3d/data/coco_data/mp_200_23_04_2020_15_37_00_coco-data'
# path = '/home/jitesh/3d/data/coco_data/sample_measure_coco_data'
# path = '/home/jitesh/3d/data/coco_data/measure_combined7'
# dest_folder_img_combined = f'{path}/img'
# dest_json_file_combined = f'{path}/json/measure-only.json'
path = '/home/pasonatech/labelme/ndds2coco/6_22/bolt_mark/type4'
dest_folder_img_combined = f'{path}'
dest_json_file_combined = f'{path}/HSR-coco.json'
dataset = COCO_Dataset.load_from_path(json_path=dest_json_file_combined,
img_dir=dest_folder_img_combined)
output = f'{path}/aug_vis'
make_dir_if_not_exists(output)
iaa = aug
# resize_save_path = 'test_resize.json'
handler_save_path = 'test_handler.json'
# if not file_exists(resize_save_path):
# resize = aug.Resize(width=500, height=500)
# resize.save_to_path(save_path=resize_save_path, overwrite=True)
# logger.info(f'Created new Resize save.')
# else:
# resize = aug.Resize.load_from_path(resize_save_path)
# logger.info(f'Loaded Resize from save.')
# if not file_exists(handler_save_path):
# handler = AugHandler(
# [
obj_type, obj_name = 'seg', 'hsr'
instance_name = '0'
ann_obj.class_name = f'{obj_type}_{obj_name}_{instance_name}'
elif ann_obj.class_name.lower() in list('abcdefghijkl'):
obj_type, obj_name = 'kpt', 'hsr'
instance_name, contained_name = '0', ann_obj.class_name
ann_obj.class_name = f'{obj_type}_{obj_name}_{instance_name}_{contained_name}'
else:
logger.error(f'Unknown ann_obj.class_name: {ann_obj.class_name}')
raise Exception
# Convert To COCO Dataset
dataset = COCO_Dataset.from_ndds(ndds_dataset=ndds_dataset,
categories=hsr_categories,
naming_rule='type_object_instance_contained',
show_pbar=True,
bbox_area_threshold=1,
allow_same_instance_for_contained=True,
color_interval=5)
dataset.move_images(dst_dir,
preserve_filenames=True,
update_img_paths=True,
overwrite=True,
show_pbar=True)
dataset.save_to_path(f'{dst_dir}/output.json', overwrite=True)
dataset.save_video(
save_path=f'{dst_dir}/preview_with_mask.avi',
fps=5,
show_details=True,
kpt_idx_offset=-1,
)
# Fix NDDS Dataset naming so that it follows convention. (This is not necessary if the NDDS dataset already follows the naming convention.)
for frame in ndds_dataset.frames:
# Fix Naming Convention
for ann_obj in frame.ndds_ann.objects:
if ann_obj.class_name.startswith('crescent'):
obj_type, obj_name = 'seg', 'crescent'
instance_name = '0'
ann_obj.class_name = f'{obj_type}_{obj_name}_{instance_name}'
# Convert To COCO Dataset
crescent_categories = COCO_Category_Handler()
crescent_categories.append(
COCO_Category(id=len(crescent_categories), name='crescent'))
dataset = COCO_Dataset.from_ndds(ndds_dataset=ndds_dataset,
categories=crescent_categories,
naming_rule='type_object_instance_contained',
show_pbar=True,
bbox_area_threshold=-1,
allow_unfound_seg=True)
dataset.move_images(dst_img_dir=dst_dir,
preserve_filenames=True,
overwrite_duplicates=False,
update_img_paths=True,
overwrite=True,
show_pbar=True)
dataset.save_to_path(f'{dst_dir}/output.json', overwrite=True)
# dataset.display_preview(show_details=True)
dataset.save_video(save_path=f'{dst_dir}/preview.mp4',
fps=5,
show_details=True)
import cv2,os
from common_utils.common_types.keypoint import Keypoint2D_List, Keypoint2D
from common_utils.common_types.segmentation import Segmentation, Polygon
import shapely
from shapely import ops
import numpy as np
from common_utils.cv_drawing_utils import draw_keypoints, cv_simple_image_viewer, draw_bbox, draw_segmentation
from imageaug import AugHandler, Augmenter as aug
save_img_path = "/home/pasonatech/detectron/detectron2/gbox/vis_image/"
dataset = COCO_Dataset.load_from_path(
json_path='/home/pasonatech/combined_cocooutput/HSR-coco.json',
img_dir='/home/pasonatech/combined_cocooutput'
#json_path='/home/pasonatech/aug_real_combine/aug_sim_com_garbage/HSR-coco.json',
#img_dir='/home/pasonatech/aug_real_combine/aug_sim_com_garbage'
)
resize_save_path = 'test_resize.json'
handler_save_path = 'test_handler.json'
if not file_exists(handler_save_path):
handler = AugHandler(
[
aug.Crop(percent=[0.2, 0.5]),
aug.Flipud(p=0.5),
aug.Superpixels()
# aug.Sharpen(alpha=[-1,0.1], lightness=[0,3])
]
)
from common_utils.file_utils import make_dir_if_not_exists, delete_all_files_in_dir
from annotation_utils.coco.structs import COCO_Dataset
dataset = COCO_Dataset.load_from_path(json_path='output.json')
# dataset.display_preview()
dump_dir = 'ann_dump'
make_dir_if_not_exists(dump_dir)
delete_all_files_in_dir(dump_dir, ask_permission=False)
dataset.info.save_to_path(f'{dump_dir}/info.json')
dataset.images.save_to_path(f'{dump_dir}/images.json')
dataset.annotations.save_to_path(f'{dump_dir}/annotations.json')
dataset.categories.save_to_path(f'{dump_dir}/categories.json')
dataset.categories.save_to_path('box_hsr_categories.json')
def infer(
path: str,
weights_path: str,
thresh: int = 0.5,
key: str = 'R',
infer_dump_dir: str = '',
model: str = 'mask_rcnn_R_50_FPN_1x',
size: int = 1024,
class_names: List[str] = ['hook'],
gt_path:
str = '/home/jitesh/3d/data/coco_data/hook_test/json/cropped_hook.json'):
# class_names=['hook', 'pole']
# class_names=['hook']
conf_thresh = 0.001
show_bbox_border = True
gt_dataset = COCO_Dataset.load_from_path(json_path=gt_path)
inferer_seg = inferer(
weights_path=weights_path,
confidence_threshold=0.1,
# num_classes=1,
# num_classes=2,
class_names=class_names,
# class_names=['hook'],
model='keypoint_rcnn_R_50_FPN_1x',
# model='faster_rcnn_X_101_32x8d_FPN_3x',
# model='faster_rcnn_R_101_FPN_3x',
# model=model,
)
inferer_seg.cfg.INPUT.MIN_SIZE_TEST = size
inferer_seg.cfg.INPUT.MAX_SIZE_TEST = size
inferer_seg.cfg.MODEL.MASK_ON = True
weights_path = '/home/jitesh/3d/data/coco_data/hook_sim_real_data7/weights/Keypoints_R_50_1x_aug_cm_seg_val_1/model_0009999.pth'
weights_path = '/home/jitesh/3d/data/coco_data/hook_sim_real_data7_0.1/weights/Keypoints_R_50_1x_aug_cm_seg_val_3/model_0009999.pth'
weights_path = '/home/jitesh/3d/data/coco_data/hook_sim_real_data7_0.1/weights/Keypoints_R_50_1x_aug_cm_seg_val_1/model_0007999.pth'
weights_path = '/home/jitesh/3d/data/coco_data/hook_sim_real_data8/weights/Keypoints_R_50_1x_aug_key_seg_val_1/model_0009999.pth'
weights_path = '/home/jitesh/3d/data/coco_data/hook_sim_real_data8/weights/Keypoints_R_50_1x_aug_key_seg_val_2/model_0004999.pth'
# inferer_key = jDetectron2KeypointInferer(
# weights_path=weights_path,
# # ref_coco_ann_path=f'/home/jitesh/3d/data/coco_data/hook_real1/json/hook.json',
# # categories_path=f'/home/jitesh/3d/data/categories/hook_infer.json',
# # categories_path=f'/home/jitesh/3d/data/categories/hook_7ckpt.json',
# categories_path=f'/home/jitesh/3d/data/categories/hook_7ckpt_pole.json',
# target_category='hook',
# model_name='keypoint_rcnn_R_50_FPN_1x',
# bbox_threshold=bbox_thresh,
# kpt_threshold=kpt_thresh,
# key_box='hook',
# )
# k_size = 1024
# inferer_key.cfg.INPUT.MIN_SIZE_TEST = k_size
# inferer_key.cfg.INPUT.MAX_SIZE_TEST = k_size
possible_modes = ['save', 'preview']
mode = 'save'
check_value(mode, valid_value_list=possible_modes)
# make_dir_if_not_exists(infer_dump_dir)
img_extensions = ['jpg', 'JPG', 'png', 'PNG']
img_pathlist = get_all_files_in_extension_list(
dir_path=f'{path}', extension_list=img_extensions)
img_pathlist.sort()
confirm_folder(infer_dump_dir, mode)
# confirm_folder(f'{infer_dump_dir}/good_seg', mode)
# confirm_folder(f'{infer_dump_dir}/good_cropped', mode)
# confirm_folder(f'{infer_dump_dir}/good', mode)
# confirm_folder(f'{infer_dump_dir}/G(>4D) P(>4D)', mode)
# confirm_folder(f'{infer_dump_dir}/G(>4D) P(<4D)', mode)
# confirm_folder(f'{infer_dump_dir}/G(<4D) P(>4D)', mode)
# confirm_folder(f'{infer_dump_dir}/G(<4D) P(<4D)', mode)
# confirm_folder(f'{infer_dump_dir}/bad', mode)
confirm_folder(f'{infer_dump_dir}/infer_key_seg', mode)
count = 0
start = datetime.now()
df = pd.DataFrame(data=[],
columns=[
'gt_d',
'pred_d',
'gt_ab',
'pred_ab',
'gt_ratio',
'pred_ratio',
'gt_ratio>4',
'pred_ratio>4',
'correct_above4d_ratio',
'incorrect_above4d_ratio',
'correct_below4d_ratio',
'incorrect_below4d_ratio',
])
# 'image_path'])
for i, img_path in enumerate(tqdm(
img_pathlist,
desc='Writing images',
)):
img_filename = get_filename(img_path)
# if not '201005_70_縮小革命PB020261.jpg' in img_path:
# continue
# if i > 19:
# continue
printj.purple(img_path)
img = cv2.imread(img_path)
result = img
# print(f'shape {img.shape}')
# cv2.imshow('i', img)
# cv2.waitKey(100000)
# continue
score_list, pred_class_list, bbox_list, pred_masks_list, pred_keypoints_list, vis_keypoints_list, kpt_confidences_list = inferer_seg.predict(
img=img)
# printj.blue(pred_masks_list)
max_hook_score = -1
max_pole_score = -1
diameter = -1
len_ab = -1
found_hook = False
found_pole = False
for score, pred_class, bbox, mask, keypoints, vis_keypoints, kpt_confidences in zip(
score_list, pred_class_list, bbox_list, pred_masks_list,
pred_keypoints_list, vis_keypoints_list, kpt_confidences_list):
if pred_class == 'pole':
found_pole = True
if max_pole_score < score:
# if True:
max_pole_score = score
diameter = compute_diameter(mask)
# result = draw_bool_mask(img=result, mask=mask, color=[
# 0, 255, 255],
# transparent=True
# )
pole_bbox_text = f'pole {str(round(score, 2))}'
pole_bbox = bbox
pole_mask = mask
# result = draw_bbox(img=result, bbox=bbox,
# text=pole_bbox_text, label_only=not show_bbox_border, label_orientation='bottom')
printj.blue(f'diameter={diameter}')
if pred_class == 'hook':
# printj.green.bold_on_yellow(score)
found_hook = True
if max_hook_score < score:
# if True:
max_hook_score = score
hook_bbox = BBox.buffer(bbox)
hook_score = round(score, 2)
hook_mask = mask
hook_keypoints = keypoints
hook_vis_keypoints = vis_keypoints
hook_kpt_confidences = kpt_confidences
# xmin, ymin, xmax, ymax = bbox.to_int().to_list()
# _xmin, _ymin, _xmax, _ymax = _bbox.to_int().to_list()
# width = _xmax-_xmin
# height = _ymax-_ymin
# scale = 0.2
# xmin = max(int(_xmin - width*scale), 0)
# xmax = min(int(_xmax + width*scale), img.shape[1])
# ymin = max(int(_ymin - height*scale), 0)
# ymax = min(int(_ymax + height*scale), img.shape[0])
# printj.red(score)
# printj.red(bbox)
# return
# img = draw_bbox(img=img, bbox=_bbox, color=[
# 0, 255, 255], thickness=2, text=f"{pred_class} {round(score, 3)}",
# label_orientation='top')
# img = draw_bbox(img=img, bbox=_bbox, color=[
# 0, 255, 255], thickness=2, text=f"{pred_class} {round(score, 3)}",
# label_orientation='bottom')
# result = draw_bool_mask(img=result, mask=mask, color=[
# 255, 255, 0],
# transparent=True
# )
# result = result
# bbox_text = str(round(score, 4))
# result = draw_bbox(img=result, bbox=bbox,
# text=bbox_text, label_only=not show_bbox_border)
bbox_label_mode = 'euler'
# result = draw_keypoints(
# img=result, keypoints=vis_keypoints, radius=2, color=[0, 0, 255],
# # keypoint_labels=kpt_labels, show_keypoints_labels=True, label_thickness=1,
# # ignore_kpt_idx=conf_idx_list
# )
kpt_labels = [
"kpt-a", "kpt-b", "kpt-cb", "kpt-c", "kpt-cd", "kpt-d",
"kpt-e"
]
kpt_skeleton = [[0, 1], [1, 2], [2, 3], [3, 4], [4, 5],
[5, 6]]
conf_idx_list = np.argwhere(
np.array(kpt_confidences) > conf_thresh).reshape(-1)
not_conf_idx_list = np.argwhere(
np.array(kpt_confidences) <= conf_thresh).reshape(
-1).astype(int)
conf_keypoints, conf_kpt_labels = np.array(vis_keypoints)[
conf_idx_list], np.array(kpt_labels)[conf_idx_list]
not_conf_keypoints, not_conf_kpt_labels = np.array(
vis_keypoints)[not_conf_idx_list], np.array(
kpt_labels)[not_conf_idx_list]
cleaned_keypoints = np.array(vis_keypoints.copy()).astype(
np.float32)
# result = draw_bool_mask(img=result, mask=mask, color=[
# 255, 255, 0],
# transparent=True
# )
# result, len_ab = draw_inference_on_hook2(img=result, cleaned_keypoints=cleaned_keypoints, kpt_labels=kpt_labels, kpt_skeleton=kpt_skeleton,
# score=score, bbox=_bbox, vis_keypoints=vis_keypoints, kpt_confidences=kpt_confidences, conf_idx_list=conf_idx_list, not_conf_idx_list=not_conf_idx_list,
# conf_keypoints=conf_keypoints, conf_kpt_labels=conf_kpt_labels, not_conf_keypoints=not_conf_keypoints, not_conf_kpt_labels=not_conf_kpt_labels,
# conf_thresh=conf_thresh, show_bbox_border=show_bbox_border, bbox_label_mode=bbox_label_mode, index_offset=0, diameter=diameter)
# result=result
# printj.green(_bbox)
# printj.green(_bbox.to_int())
# printj.green(_bbox.to_int().to_list())
printj.green.on_white(max_hook_score)
if found_pole:
result = draw_bool_mask(img=result,
mask=pole_mask,
color=[0, 255, 255],
transparent=True)
result = draw_bbox(img=result,
bbox=pole_bbox,
text=pole_bbox_text,
label_only=not show_bbox_border,
label_orientation='top')
result = draw_bbox(img=result,
bbox=pole_bbox,
text=pole_bbox_text,
label_only=not show_bbox_border,
label_orientation='bottom')
if found_hook:
result = draw_bool_mask(img=result,
mask=hook_mask,
color=[255, 255, 0],
transparent=True)
result, len_ab = draw_inference_on_hook2(
img=result,
cleaned_keypoints=cleaned_keypoints,
kpt_labels=kpt_labels,
kpt_skeleton=kpt_skeleton,
score=hook_score,
bbox=hook_bbox,
vis_keypoints=hook_vis_keypoints,
kpt_confidences=hook_kpt_confidences,
conf_idx_list=conf_idx_list,
not_conf_idx_list=not_conf_idx_list,
conf_keypoints=conf_keypoints,
conf_kpt_labels=conf_kpt_labels,
not_conf_keypoints=not_conf_keypoints,
not_conf_kpt_labels=not_conf_kpt_labels,
conf_thresh=conf_thresh,
show_bbox_border=show_bbox_border,
bbox_label_mode=bbox_label_mode,
index_offset=0,
diameter=diameter)
printj.purple(len_ab)
if len_ab == 0:
printj.green(keypoints)
result = draw_info_box(result, len_ab, diameter)
# img: np.ndarray, cleaned_keypoints, kpt_labels: List[str], kpt_skeleton: List[list],
# score: float, bbox: BBox, vis_keypoints: list, kpt_confidences: list, conf_idx_list: list, not_conf_idx_list: list,
# conf_keypoints, conf_kpt_labels, not_conf_keypoints, not_conf_kpt_labels,
# conf_thresh: float = 0.3, show_bbox_border: bool = False, bbox_label_mode: str = 'euler', index_offset: int = 0, diameter=1
# cv2.imshow('i', result)
# # cv2.imwrite('i', result)
# cv2.waitKey(10000)
# quit_flag = cv_simple_image_viewer(img=result, preview_width=1000)
# if quit_flag:
# break
# cv2.imwrite(f"{infer_dump_dir}/good_seg/{img_filename}", result)
cv2.imwrite(f"{infer_dump_dir}/infer_key_seg/{img_filename}", result)
from annotation_utils.coco.structs import COCO_Dataset
dataset = COCO_Dataset.load_from_path(
json_path=
'/home/clayton/workspace/prj/data_keep/data/toyota/dataset/sim/20200228/28_02_2020_11_18_30_coco-data/HSR-coco.json', # Modify this path
check_paths=False)
dataset.print_handler_lengths()
from annotation_utils.coco.structs import COCO_Dataset
# dataset = COCO_Dataset.load_from_path(
# json_path='/home/clayton/workspace/prj/data_keep/data/toyota/dataset/sim/20200228/28_02_2020_11_18_30_coco-data/HSR-coco.json',
# img_dir='/home/clayton/workspace/prj/data_keep/data/toyota/dataset/sim/20200228/28_02_2020_11_18_30_coco-data'
# )
dataset = COCO_Dataset.combine_from_config(
'/home/clayton/workspace/prj/data_keep/data/toyota/dataset/config/json/box_hsr_kpt_real.json'
)
dataset.images.sort(attr_name='file_name')
dataset.save_video(save_path='dataset.mp4',
show_annotations=True,
fps=5,
show_preview=True,
kpt_idx_offset=-1,
overwrite=True,
show_bbox=False)
from annotation_utils.coco.structs import COCO_Dataset
dataset = COCO_Dataset.load_from_path('measure_total_test.json')
dataset.remove_all_categories_except(['measure'])
dataset.display_preview(show_details=True)
from annotation_utils.coco.structs import COCO_Dataset
from logger import logger
from common_utils.image_utils import concat_n_images
from common_utils.cv_drawing_utils import cv_simple_image_viewer
from common_utils.file_utils import file_exists
import cv2
from common_utils.common_types.keypoint import Keypoint2D_List, Keypoint2D
from imageaug import AugHandler, Augmenter as aug
dataset = COCO_Dataset.load_from_path(
json_path=
'/Users/darwinharianto/Desktop/hayashida/Unreal/18_03_2020_18_03_10_coco-data/HSR-coco.json',
img_dir=
'/Users/darwinharianto/Desktop/hayashida/Unreal/18_03_2020_18_03_10_coco-data'
)
resize_save_path = 'test_resize.json'
handler_save_path = 'test_handler.json'
if not file_exists(resize_save_path):
resize = aug.Resize(width=500, height=500)
resize.save_to_path(save_path=resize_save_path, overwrite=True)
logger.info(f'Created new Resize save.')
else:
resize = aug.Resize.load_from_path(resize_save_path)
logger.info(f'Loaded Resize from save.')
if not file_exists(handler_save_path):
handler = AugHandler([
aug.Crop(percent=[0.2, 0.5]),
aug.Flipud(p=0.5),
aug.Superpixels()
from annotation_utils.labelme.structs import LabelmeAnnotationHandler
from annotation_utils.coco.structs import COCO_Dataset, COCO_Category_Handler
# Path Variables
img_dir = ''
json_dir = ''
categories_conf_path = ''
# Load Labelme Annotation Handler
labelme_handler = LabelmeAnnotationHandler.load_from_dir(load_dir=json_dir)
# Load COCO Dataset from Labelme Annotation Handler
coco_dataset = COCO_Dataset.from_labelme(
labelme_handler=labelme_handler,
categories=COCO_Category_Handler.load_from_path(categories_conf_path),
img_dir=img_dir
)
# Save COCO Dataset to a file
coco_dataset.save_to_path(save_path='', overwrite=True)
# Preview COCO Dataset
coco_dataset.display_preview(show_details=True)
coco_dataset.save_visualization(show_details=True)
# coco_dataset.save_video(show_details=True)
logger.error(f'ann_obj.class_name: {ann_obj.class_name}')
# raise Exception
# Delete Duplicate Objects
frame.ndds_ann.objects.delete_duplicates(verbose=True,
verbose_ref=frame.img_path)
# ndds_dataset.save_to_path(save_path=f'{coco_data_dir}/hook_fixed_ndds.json', overwrite=True)
# Convert To COCO Dataset
dataset = COCO_Dataset.from_ndds(
ndds_dataset=ndds_dataset,
# categories=COCO_Category_Handler.load_from_path(f'/home/jitesh/3d/data/categories/hook_7ckpt.json'),
categories=COCO_Category_Handler.load_from_path(
f'/home/jitesh/3d/data/categories/tropicana.json'),
naming_rule='type_object_instance_contained',
ignore_unspecified_categories=True,
show_pbar=True,
bbox_area_threshold=1,
default_visibility_threshold=-1,
allow_unfound_seg=True,
)
make_dir_if_not_exists(coco_data_dir)
img_path = f'{coco_data_dir}/img'
make_dir_if_not_exists(coco_data_dir)
ann_dir = f'{coco_data_dir}/json'
make_dir_if_not_exists(ann_dir)
dataset.move_images(dst_img_dir=img_path,
preserve_filenames=False,
update_img_paths=True,
overwrite=True,
show_pbar=True)
from annotation_utils.coco.structs import COCO_Dataset, \
COCO_License, COCO_Image, COCO_Annotation, COCO_Category
from common_utils.common_types.bbox import BBox
from logger import logger
dataset = COCO_Dataset.new(description='Test')
dataset.categories.append(
COCO_Category(id=len(dataset.categories),
supercategory='test_category',
name='category_a'))
dataset.categories.append(
COCO_Category(id=len(dataset.categories),
supercategory='test_category',
name='category_b'))
dataset.categories.append(
COCO_Category(id=len(dataset.categories),
supercategory='test_category',
name='category_c'))
for i in range(10):
dataset.licenses.append(
COCO_License(url=f'test_license_{i}',
name=f'Test License {i}',
id=len(dataset.licenses)))
for i in range(20):
dataset.images.append(
COCO_Image(license_id=i % len(dataset.licenses),
file_name=f'{i}.jpg',
coco_url=f'/path/to/{i}.jpg',
height=500,
width=500,
bolt_roi_categories.append(
COCO_Category(
id=len(bolt_roi_categories),
name='bolt-roi'
)
)
print(f"Bolt_roi_categories :{bolt_roi_categories}")
bolt_roi_dataset = COCO_Dataset.from_ndds(
ndds_dataset=ndds_dataset,
categories=bolt_roi_categories,
naming_rule='type_object_instance_contained', delimiter='_',
ignore_unspecified_categories=True,
show_pbar=True,
bbox_area_threshold=1,
default_visibility_threshold=0.01,
visibility_threshold_dict={'bolt-roi': 0.01},
allow_unfound_seg=False,
class_merge_map={
'seg_mark-inner_0': 'seg_bolt-roi_0',
'seg_mark-middle_0': 'seg_bolt-roi_0',
'seg_mark-outer_0': 'seg_bolt-roi_0'
}
)
bolt_roi_dst_dir = f'{target_dst_dir}/bolt_roi'
make_dir_if_not_exists(bolt_roi_dst_dir)
bolt_roi_dataset.move_images(
dst_img_dir=bolt_roi_dst_dir,
preserve_filenames=True, overwrite_duplicates=False, update_img_paths=True, overwrite=True,
show_pbar=True
)
bolt_roi_dataset.save_to_path(f'{bolt_roi_dst_dir}/output.json', overwrite=True)
from annotation_utils.labelme.structs import LabelmeAnnotationHandler
from annotation_utils.coco.structs import COCO_Dataset, COCO_Category_Handler
img_dir = '/home/clayton/workspace/prj/data_keep/data/toyota/dataset/real/phone_videos/new/sampled_data/VID_20200217_161043/img'
json_dir = '/home/clayton/workspace/prj/data_keep/data/toyota/dataset/real/phone_videos/new/sampled_data/VID_20200217_161043/json'
labelme_handler = LabelmeAnnotationHandler.load_from_dir(load_dir=json_dir)
coco_dataset = COCO_Dataset.from_labelme(
labelme_handler=labelme_handler,
categories=COCO_Category_Handler.load_from_path(
'/home/clayton/workspace/prj/data_keep/data/toyota/dataset/config/categories/hsr_categories.json'
),
img_dir=img_dir)
coco_dataset.save_to_path(save_path='output.json', overwrite=True)
test_coco_dataset = COCO_Dataset.load_from_path(json_path='output.json')
test_coco_dataset.display_preview(kpt_idx_offset=-1)
# logger.cyan(info1.contributor)
# logger.purple(f'info:\n{info}')
# logger.purple(f'info0:\n{info0}')
# logger.purple(f'info.to_dict():\n{info.to_dict()}')
info2 = COCO_Info.from_dict({
'description': 'This is a test',
'url': 'https://test/url.com',
'version': '1.0',
'year': '2020',
'contributor': 'Clayton',
'date_created': '2020/03/10'
})
logger.purple(f'info2:\n{info2}')
info2.save_to_path('info.json', overwrite=True)
from annotation_utils.coco.structs import COCO_Dataset
dataset = COCO_Dataset.combine_from_config(
config_path=
'/home/clayton/workspace/prj/data_keep/data/toyota/dataset/config/yaml/box_hsr_kpt_sim.yaml',
img_sort_attr_name='file_name',
show_pbar=True)
# dataset.display_preview(kpt_idx_offset=-1)
coco_ann = dataset.annotations[123]
logger.purple(coco_ann)
from common_utils.common_types.keypoint import Keypoint3D_List
coco_ann.keypoints_3d = Keypoint3D_List()
logger.purple(f'coco_ann.to_dict():\n{coco_ann.to_dict()}')
from annotation_utils.linemod.objects import Linemod_Dataset, LinemodCamera
from annotation_utils.coco.structs import COCO_Dataset
inferer = PVNetInferer(
# weight_path='/home/clayton/workspace/git/clean-pvnet/data/model/pvnet/custom/99.pth',
# weight_path='/home/clayton/workspace/prj/data_keep/data/misc_dataset/darwin_weights/194.pth',
weight_path=
'/home/clayton/workspace/prj/data_keep/data/weights/pvnet_hsr/20201119/499.pth'
)
# img_dir = '/home/clayton/workspace/git/pvnet-rendering/test/renders1'
img_dir = '/home/clayton/workspace/prj/data_keep/data/misc_dataset/darwin_datasets/nihonbashi2/organized'
linemod_dataset = Linemod_Dataset.load_from_path(f'{img_dir}/train.json')
coco_dataset = COCO_Dataset.load_from_path(
json_path=
'/home/clayton/workspace/prj/data_keep/data/toyota/from_toyota/20201017/20201017_robot_camera/combined/output.json',
img_dir=
'/home/clayton/workspace/prj/data_keep/data/toyota/from_toyota/20201017/20201017_robot_camera/combined'
)
linemod_ann_sample = linemod_dataset.annotations[0]
kpt_3d = linemod_ann_sample.fps_3d.copy()
kpt_3d.append(linemod_ann_sample.center_3d)
corner_3d = linemod_ann_sample.corner_3d
K = linemod_ann_sample.K
linemod_image_sample = linemod_dataset.images[0]
dsize = (linemod_image_sample.width, linemod_image_sample.height)
inferer.infer_coco_dataset(dataset=coco_dataset,
kpt_3d=kpt_3d,
corner_3d=corner_3d,
K=K,
blackout=True,
'pvnet-darwin20210105-epoch599',
'pvnet-darwin20210105-epoch699',
'pvnet-darwin20210105-epoch799',
]
test_root_dir = '/home/clayton/workspace/prj/data_keep/data/toyota/from_toyota/20201017/20201017_robot_camera'
csv_paths = recursively_get_all_filepaths_of_extension(test_root_dir,
extension='csv')
test_names, datasets = [], []
for csv_path in csv_paths:
test_name = get_rootname_from_path(csv_path)
img_dir = f'{get_dirpath_from_filepath(csv_path)}/images'
assert dir_exists(img_dir), f"Couldn't find image directory: {img_dir}"
ann_path = f'{img_dir}/output.json'
if not file_exists(ann_path):
continue
dataset = COCO_Dataset.load_from_path(ann_path, img_dir=img_dir)
test_names.append(test_name)
datasets.append(dataset)
linemod_dataset = Linemod_Dataset.load_from_path(
f'/home/clayton/workspace/prj/data/misc_dataset/darwin_datasets/coco2linemod/darwin20210105_blackout/train.json'
)
linemod_ann_sample = linemod_dataset.annotations[0]
kpt_3d = linemod_ann_sample.fps_3d.copy()
kpt_3d.append(linemod_ann_sample.center_3d)
corner_3d = linemod_ann_sample.corner_3d
# K = linemod_ann_sample.K
K = np.array([
517.799858, 0.000000, 303.876287, 0.000000, 514.807834, 238.157119,
0.000000, 0.000000, 1.000000
]).reshape(3, 3)
from annotation_utils.coco.structs import COCO_Dataset
dataset = COCO_Dataset.load_from_path(
'/home/clayton/workspace/prj/data_keep/data/dataset/bird/img/output.json',
img_dir='/home/clayton/workspace/prj/data_keep/data/dataset/bird/img')
# split_image_handlers = dataset.images.split(ratio=[1, 3, 1], shuffle=True)
# for image_handler in split_image_handlers:
# print(f'len(image_handler): {len(image_handler)}')
# print(f'\tfilenames: {[coco_image.file_name for coco_image in image_handler]}')
parts = dataset.split_into_parts(ratio=[1, 2, 3], shuffle=True)
part_count = -1
for part in parts:
part_count += 1
print(f'len(part.images): {len(part.images)}')
part.save_video(save_path=f'part{part_count}.avi',
fps=5,
show_details=True)
from annotation_utils.coco.structs import COCO_Dataset
dataset = COCO_Dataset.load_from_path(
json_path=
'/home/clayton/workspace/prj/data_keep/data/toyota/dataset/sim/20200228/28_02_2020_11_18_30_coco-data/HSR-coco.json',
img_dir=
'/home/clayton/workspace/prj/data_keep/data/toyota/dataset/sim/20200228/28_02_2020_11_18_30_coco-data'
)
dataset.images.sort(attr_name='file_name')
dataset.save_visualization(save_dir='test_vis',
show_preview=True,
kpt_idx_offset=-1,
overwrite=True)
inferer = PVNetInferer(
weight_path=
'/home/clayton/workspace/prj/data_keep/data/weights/pvnet_hsr/20201119/499.pth'
)
linemod_dataset = Linemod_Dataset.load_from_path(
'/home/clayton/workspace/prj/data_keep/data/misc_dataset/clayton_datasets/combined/output.json'
)
kpt_3d, corner_3d, K = linemod_dataset.sample_3d_hyperparams(
idx=0) # Should be constant
dsize = linemod_dataset.sample_dsize(idx=0) # Assuming constant resolution
# Define test dataset that you want to run inference on
img_dir = '/home/clayton/workspace/prj/data_keep/data/misc_dataset/darwin_datasets/darwin20201209/coco_data'
ann_path = f'{img_dir}/coco_annotations.json'
test_dataset = COCO_Dataset.load_from_path(ann_path,
img_dir=img_dir,
strict=False)
test_dataset, _ = test_dataset.split_into_parts(
ratio=[100, len(test_dataset.images) - 100], shuffle=True)
# Define a StreamWriter for creating a preview window and saving video
stream_writer = StreamWriter(show_preview=True,
video_save_path='example_pnp_inference.avi')
draw_settings = PnpDrawSettings()
draw_settings.direction_line_color = (255, 255, 0)
# Initialize Inference Dump Data Handler
pred_list = PVNetFrameResultList()
# Inference Loop
pbar = tqdm(total=len(test_dataset.images), unit='image(s)', leave=True)
categories.append( # Simple Keypoint Example
COCO_Category.from_label_skeleton(
id=len(categories),
supercategory='pet',
name='cat',
label_skeleton=[
['left_eye', 'right_eye'],
['mouth_left', 'mouth_center'], ['mouth_center', 'mouth_right']
]
)
)
for name in ['duck', 'sparrow', 'pigion']:
categories.append( # Simple Non-Keypoint Example
COCO_Category(
id=len(categories),
supercategory='bird',
name=name
)
)
categories.save_to_path('categories_example.json')
else: # Or load from an existing categories json
categories = COCO_Category_Handler.load_from_path('categories_example.json')
# Convert To COCO
coco_dataset = COCO_Dataset.from_labelme(
labelme_handler=labelme_handler,
categories=categories,
img_dir=img_dir
)
coco_dataset.save_to_path(save_path='converted_coco.json', overwrite=True)
coco_dataset.display_preview(show_details=True) # Optional: Preview your resulting dataset.
for ann_obj in frame.ndds_ann.objects:
if ann_obj.class_name.startswith('hsr'):
obj_type, obj_name = 'seg', 'hsr'
instance_name = ann_obj.class_name.replace('hsr', '')
ann_obj.class_name = f'{obj_type}_{obj_name}_{instance_name}'
elif ann_obj.class_name.startswith('point'):
obj_type, obj_name = 'kpt', 'hsr'
temp = ann_obj.class_name.replace('point', '')
instance_name, contained_name = temp[1], temp[0]
ann_obj.class_name = f'{obj_type}_{obj_name}_{instance_name}_{contained_name}'
else:
logger.error(f'ann_obj.class_name: {ann_obj.class_name}')
raise Exception
# Delete Duplicate Objects
frame.ndds_ann.objects.delete_duplicates(verbose=True,
verbose_ref=frame.img_path)
ndds_dataset.save_to_path(save_path='hsr_fixed_ndds.json', overwrite=True)
# Convert To COCO Dataset
dataset = COCO_Dataset.from_ndds(
ndds_dataset=ndds_dataset,
categories=COCO_Category_Handler.load_from_path(
'/home/clayton/workspace/prj/data_keep/data/ndds/categories/hsr.json'),
naming_rule='type_object_instance_contained',
show_pbar=True,
bbox_area_threshold=1)
dataset.save_to_path('hsr_ndds2coco_test.json', overwrite=True)
dataset.display_preview(show_details=True)
from annotation_utils.coco.structs import COCO_Dataset
dataset = COCO_Dataset.load_from_path('measure_coco/measure/output.json')
dataset.save_video(
save_path='merged_mask_measure_viz.mp4',
show_details=True,
fps=3
)
from annotation_utils.coco.structs import COCO_Dataset
dataset = COCO_Dataset.load_from_path(
json_path=
'/home/clayton/Downloads/screw1_mark-100-_coco-data/screw-coco.json',
img_dir='/home/clayton/Downloads/screw1_mark-100-_coco-data',
strict=False)
screw_dataset, mark_dataset = dataset.copy(), dataset.copy()
screw_dataset.remove_categories_by_name(category_names=['mark'])
mark_dataset.remove_categories_by_name(category_names=['screw'])
screw_dataset.save_to_path(save_path='screw.json', overwrite=True)
mark_dataset.save_to_path(save_path='mark.json', overwrite=True)
screw_dataset.display_preview(show_details=True, show_seg=False)
mark_dataset.display_preview(show_details=True, )