Upload 2 files

cityscape.py

@@ -0,0 +1,130 @@
+import pandas as pd
+from huggingface_hub import hf_hub_url   # hf_hub_url is used to construct the URL of a file from the given information.
+
+import datasets
+import os
+
+#_VERSION = datasets.Version("0.0.2")
+
+_DESCRIPTION = "TODO"
+_HOMEPAGE = "TODO"
+_LICENSE = "TODO"
+_CITATION = "TODO"
+
+_FEATURES = datasets.Features(
+    {
+        "image": datasets.Image(),
+        "conditioning_image": datasets.Image(),
+        "text": datasets.Value("string"),
+    },
+)
+
+METADATA_URL = hf_hub_url(
+    "MoaazId/cityscape",
+    filename="train.jsonl",
+    repo_type="dataset",
+)
+
+IMAGES_URL = hf_hub_url(
+    "MoaazId/cityscape",
+    filename="leftImg8bit.zip/train/"+f"{city}", #M
+    repo_type="dataset",
+)
+
+CONDITIONING_IMAGES_URL = hf_hub_url(
+    "MoaazId/cityscape",
+    filename="gtCoarse.zip/train/"+f"{city}", #M
+    repo_type="dataset",
+)
+
+_DEFAULT_CONFIG = datasets.BuilderConfig(name="default")
+
+
+class Cityscape(datasets.GeneratorBasedBuilder):
+    BUILDER_CONFIGS = [_DEFAULT_CONFIG]
+    DEFAULT_CONFIG_NAME = "default"
+
+    def _info(self):
+        return datasets.DatasetInfo(
+            description=_DESCRIPTION,
+            features=_FEATURES,
+            supervised_keys=None,
+            homepage=_HOMEPAGE,
+            license=_LICENSE,
+            citation=_CITATION,
+        )
+
+    def _split_generators(self, dl_manager):
+        metadata_path = dl_manager.download(METADATA_URL)
+        images_dir = dl_manager.download_and_extract(IMAGES_URL)
+        conditioning_images_dir = dl_manager.download_and_extract(
+            CONDITIONING_IMAGES_URL
+        )
+	
+	
+	# Iterate through the city folders in the image directory
+	for city_folder in os.listdir(images_dir):
+		city_dir = os.path.join(images_dir, city_folder)
+    
+    		# Iterate through image files in the city directory
+    		for image_filename in os.listdir(city_dir):
+        		# Extract relevant information from the image filename
+        		parts = image_filename.split("_")
+        		city = parts[0]
+        		seq = parts[1]
+        		frame = parts[2]
+        
+        		# Construct the paths to the image and conditioning image
+        		image_path = os.path.join(city_dir, image_filename)
+        		conditioning_image_filename = f"{city}_{seq}_{frame}_gtCoarse_color.png"
+        		conditioning_image_path = os.path.join(conditioning_dir, city_folder+"/"+conditioning_image_filename)
+        
+        		# Create a dataset entry as a dictionary
+        		entry = {
+            			"text": "A view to a street from a car's front window",
+            			"image": image_path,
+            			"conditioning_image": conditioning_image_path,
+        		}
+        
+	
+	
+	
+        return [
+            datasets.SplitGenerator(
+                name=datasets.Split.TRAIN,
+                # These kwargs will be passed to _generate_examples
+                gen_kwargs={
+                    "metadata_path": metadata_path,
+                    "images_dir": images_dir,
+                    "conditioning_images_dir": conditioning_images_dir,
+                },
+            ),
+        ]
+
+    def _generate_examples(self, metadata_path, images_dir, conditioning_images_dir):
+        metadata = pd.read_json(metadata_path, lines=True)
+
+        for _, row in metadata.iterrows():
+            text = row["text"]
+
+            image_path = row["image"]
+            image_path = os.path.join(images_dir, image_path)
+            image = open(image_path, "rb").read()
+
+            conditioning_image_path = row["conditioning_image"]
+            conditioning_image_path = os.path.join(
+                conditioning_images_dir, row["conditioning_image"]
+            )
+            conditioning_image = open(conditioning_image_path, "rb").read()
+
+            yield row["image"], {
+                "text": text,
+                "image": {
+                    "path": image_path,
+                    "bytes": image,
+                },
+                "conditioning_image": {
+                    "path": conditioning_image_path,
+                    "bytes": conditioning_image,
+                },
+            }

create_jsonl_file.py

@@ -0,0 +1,53 @@
+import os
+import json
+
+# Define the directory where your dataset images are located
+#dataset_directory = "cityscape_dataset"
+
+# Define the directories for images and conditioning images
+#image_dir = os.path.join(dataset_directory, "leftImg8bit/train")
+#conditioning_dir = os.path.join(dataset_directory, "gtCoarse/train")
+
+image_dir = "leftImg8bit/train"
+conditioning_dir = "gtCoarse/train"
+
+# Define the output JSONL filename
+jsonl_filename = "train.jsonl"
+
+# Initialize an empty list to store the dataset entries
+dataset_entries = []
+
+# Iterate through the city folders in the image directory
+for city_folder in os.listdir(image_dir):
+    city_dir = os.path.join(image_dir, city_folder)
+    
+    # Iterate through image files in the city directory
+    for image_filename in os.listdir(city_dir):
+        # Extract relevant information from the image filename
+        parts = image_filename.split("_")
+        city = parts[0]
+        seq = parts[1]
+        frame = parts[2]
+        
+        # Construct the paths to the image and conditioning image
+        image_path = os.path.join(city_dir, image_filename)
+        conditioning_image_filename = f"{city}_{seq}_{frame}_gtCoarse_color.png"
+        conditioning_image_path = os.path.join(conditioning_dir, city_folder+"/"+conditioning_image_filename)
+        
+        # Create a dataset entry as a dictionary
+        entry = {
+            "text": "A view to a street from a car's front window",
+            "image": image_path,
+            "conditioning_image": conditioning_image_path,
+        }
+        
+        # Append the entry dictionary to the list
+        dataset_entries.append(entry)
+
+# Open the JSONL file for writing
+with open(jsonl_filename, "w") as jsonl_file:
+    # Write each entry as a JSON string followed by a newline character
+    for entry in dataset_entries:
+        jsonl_file.write(json.dumps(entry) + "\n")
+
+print(f"JSON Lines file '{jsonl_filename}' has been created.")