# coding=utf-8 # Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. """VALERIE22 dataset""" import os import json import glob import datasets _HOMEPAGE = "https://huggingface.co/datasets/Intel/VALERIE22" _LICENSE = "Creative Commons — CC0 1.0 Universal" _CITATION = """\ tba """ _DESCRIPTION = """\ The VALERIE22 dataset was generated with the VALERIE procedural tools pipeline providing a photorealistic sensor simulation rendered from automatically synthesized scenes. The dataset provides a uniquely rich set of metadata, allowing extraction of specific scene and semantic features (like pixel-accurate occlusion rates, positions in the scene and distance + angle to the camera). This enables a multitude of possible tests on the data and we hope to stimulate research on understanding performance of DNNs. """ _REPO = "https://huggingface.co/datasets/Intel/VALERIE22/resolve/main" _SEQUENCES = { "train": ["intel_results_sequence_0057.zip", "intel_results_sequence_0058.zip", "intel_results_sequence_0059.zip", "intel_results_sequence_0060.zip", "intel_results_sequence_0062_part1.zip", "intel_results_sequence_0062_part2.zip"], "validation":["intel_results_sequence_0062_part1.zip", "intel_results_sequence_0062_part2.zip"], "test":["intel_results_sequence_0062_part1.zip", "intel_results_sequence_0062_part2.zip"] } _URLS = { "train": [f"{_REPO}/data/{sequence}" for sequence in _SEQUENCES["train"]], "validation": [f"{_REPO}/data/{sequence}" for sequence in _SEQUENCES["validation"]], "test": [f"{_REPO}/data/{sequence}" for sequence in _SEQUENCES["test"]] } class VALERIE22(datasets.GeneratorBasedBuilder): """VALERIE22 dataset.""" VERSION = datasets.Version("1.0.0") def _info(self): return datasets.DatasetInfo( description=_DESCRIPTION, features=datasets.Features( { "image": datasets.Image(), "image_distorted": datasets.Image(), "persons_png": datasets.Sequence( { "bbox": datasets.Sequence(datasets.Value("float32"), length=4), "bbox_vis": datasets.Sequence(datasets.Value("float32"), length=4), "occlusion": datasets.Value("float32"), "distance": datasets.Value("float32"), "v_x": datasets.Value("float32"), "v_y": datasets.Value("float32"), "truncated": datasets.Value("bool"), "total_pixels_object": datasets.Value("float32"), "total_visible_pixels_object": datasets.Value("float32"), "contrast_rgb_full": datasets.Value("float32"), "contrast_edge": datasets.Value("float32"), "contrast_rgb": datasets.Value("float32"), "luminance": datasets.Value("float32"), "perceived_lightness": datasets.Value("float32"), "3dbbox": datasets.Sequence(datasets.Value("float32"), length=6) # 3center, 3 size } ), "persons_png_distorted": datasets.Sequence( { "bbox": datasets.Sequence(datasets.Value("float32"), length=4), "bbox_vis": datasets.Sequence(datasets.Value("float32"), length=4), "occlusion": datasets.Value("float32"), "distance": datasets.Value("float32"), "v_x": datasets.Value("float32"), "v_y": datasets.Value("float32"), "truncated": datasets.Value("bool"), "total_pixels_object": datasets.Value("float32"), "total_visible_pixels_object": datasets.Value("float32"), "contrast_rgb_full": datasets.Value("float32"), "contrast_edge": datasets.Value("float32"), "contrast_rgb": datasets.Value("float32"), "luminance": datasets.Value("float32"), "perceived_lightness": datasets.Value("float32"), "3dbbox": datasets.Sequence(datasets.Value("float32"), length=6) # 3center, 3 size } ), "semantic_group_segmentation": datasets.Image(), "semantic_instance_segmentation": datasets.Image() } ), supervised_keys=None, homepage=_HOMEPAGE, license=_LICENSE, citation=_CITATION, ) def _split_generators(self, dl_manager): data_dir = dl_manager.download_and_extract(_URLS) return [ datasets.SplitGenerator( name=datasets.Split.TRAIN, gen_kwargs={ "split": "train", "data_dirs": data_dir["train"], }, ), datasets.SplitGenerator( name=datasets.Split.TEST, gen_kwargs={ "split": "test", "data_dirs": data_dir["test"], }, ), datasets.SplitGenerator( name=datasets.Split.VALIDATION, gen_kwargs={ "split": "validation", "data_dirs": data_dir["validation"], }, ), ] def _generate_examples(self, split, data_dirs): sequence_dirs = [] for data_dir, sequence in zip(data_dirs, _SEQUENCES[split]): sequence = sequence.replace(".zip","") if "_part1" in sequence: sequence = sequence.replace("_part1","") if "_part2" in sequence: sequence_0062_part2_dir = os.path.join(data_dir, sequence.replace("_part2","_b")) continue sequence_dirs.append(os.path.join(data_dir, sequence)) idx = 0 for sequence_dir in sequence_dirs: for filename in glob.glob(os.path.join(os.path.join(sequence_dir, "sensor/camera/left/png"), "*.png")): # image_file_path image_file_path = filename # image_distorted_file_path if "_0062" in sequence_dir: image_distorted_file_path = os.path.join(sequence_0062_part2_dir, "sensor/camera/left/png_distorted/", os.path.basename(filename)) else: image_distorted_file_path = filename.replace("/png/", "/png_distorted/") #persons_png persons_png_path = filename.replace("sensor/camera/left/png/", "ground-truth/2d-bounding-box_json/") #persons_distorted_png persons_distorted_png_path = filename.replace("sensor/camera/left/png/", "ground-truth/2d-bounding-box_json_png_distorted/") #semantic_group_segmentation_file_path semantic_group_segmentation_file_path = filename.replace("sensor/camera/left/png/", "ground-truth/semantic-group-segmentation_png/") # semantic_instance_segmentation_file_path semantic_instance_segmentation_file_path = filename.replace("sensor/camera/left/png/", "ground-truth/semantic-instance-segmentation_png/") # check if all gt files are available if not (os.path.isfile(image_file_path) and os.path.isfile(image_distorted_file_path) and os.path.isfile(persons_png_path.replace(".png",".json")) and os.path.isfile(persons_distorted_png_path.replace(".png",".json")) and os.path.isfile(semantic_group_segmentation_file_path) and os.path.isfile(semantic_instance_segmentation_file_path)): continue with open(persons_png_path.replace(".png",".json"), 'r') as json_file: bb_person_json = json.load(json_file) with open(persons_distorted_png_path.replace(".png",".json"), 'r') as json_file: bb_person_distorted_json = json.load(json_file) threed_bb_person_path = filename.replace("sensor/camera/left/png/", "ground-truth/3d-bounding-box_json/") with open(os.path.join(threed_bb_person_path.replace(".png",".json")), 'r') as json_file: threed_bb_person_distorted_json = json.load(json_file) persons_png = [] persons_png_distorted = [] for key in bb_person_json: persons_png.append( { "bbox": [bb_person_json[key]["bb"]["c_x"], bb_person_json[key]["bb"]["c_y"], bb_person_json[key]["bb"]["w"], bb_person_json[key]["bb"]["h"]], "bbox_vis": [bb_person_json[key]["bb_vis"]["c_x"], bb_person_json[key]["bb_vis"]["c_y"], bb_person_json[key]["bb_vis"]["w"], bb_person_json[key]["bb_vis"]["h"]], "occlusion": bb_person_json[key]["occlusion"], "distance": bb_person_json[key]["distance"], "v_x": bb_person_json[key]["v_x"], "v_y": bb_person_json[key]["v_y"], "truncated": bb_person_json[key]["truncated"], "total_pixels_object": bb_person_json[key]["total_pixels_object"], "total_visible_pixels_object": bb_person_json[key]["total_visible_pixels_object"], "contrast_rgb_full": bb_person_json[key]["contrast_rgb_full"], "contrast_edge": bb_person_json[key]["contrast_edge"], "contrast_rgb": bb_person_json[key]["contrast_rgb"], "luminance": bb_person_json[key]["luminance"], "perceived_lightness": bb_person_json[key]["perceived_lightness"], "3dbbox": [threed_bb_person_distorted_json[key]["center"][0], threed_bb_person_distorted_json[key]["center"][1], threed_bb_person_distorted_json[key]["center"][2], threed_bb_person_distorted_json[key]["size"][0], threed_bb_person_distorted_json[key]["size"][1], threed_bb_person_distorted_json[key]["size"][2]] # 3center, 3 size } ) persons_png_distorted.append( { "bbox": [bb_person_distorted_json[key]["bb"]["c_x"], bb_person_distorted_json[key]["bb"]["c_y"], bb_person_distorted_json[key]["bb"]["w"], bb_person_distorted_json[key]["bb"]["h"]], "bbox_vis": [bb_person_distorted_json[key]["bb_vis"]["c_x"], bb_person_distorted_json[key]["bb_vis"]["c_y"], bb_person_distorted_json[key]["bb_vis"]["w"], bb_person_distorted_json[key]["bb_vis"]["h"]], "occlusion": bb_person_distorted_json[key]["occlusion"], "distance": bb_person_distorted_json[key]["distance"], "v_x": bb_person_distorted_json[key]["v_x"], "v_y": bb_person_distorted_json[key]["v_y"], "truncated": bb_person_distorted_json[key]["truncated"], "total_pixels_object": bb_person_distorted_json[key]["total_pixels_object"], "total_visible_pixels_object": bb_person_distorted_json[key]["total_visible_pixels_object"], "contrast_rgb_full": bb_person_distorted_json[key]["contrast_rgb_full"], "contrast_edge": bb_person_distorted_json[key]["contrast_edge"], "contrast_rgb": bb_person_distorted_json[key]["contrast_rgb"], "luminance": bb_person_distorted_json[key]["luminance"], "perceived_lightness": bb_person_distorted_json[key]["perceived_lightness"], "3dbbox": [threed_bb_person_distorted_json[key]["center"][0], threed_bb_person_distorted_json[key]["center"][1], threed_bb_person_distorted_json[key]["center"][2], threed_bb_person_distorted_json[key]["size"][0], threed_bb_person_distorted_json[key]["size"][1], threed_bb_person_distorted_json[key]["size"][2]] # 3center, 3 size } ) yield idx, {"image": image_file_path, "image_distorted": image_distorted_file_path, "persons_png": persons_png, "persons_png_distorted":persons_png_distorted, "semantic_group_segmentation": semantic_group_segmentation_file_path, "semantic_instance_segmentation": semantic_instance_segmentation_file_path} idx += 1