feat: script

facial-emotion-recognition-dataset.py

@@ -9,32 +9,24 @@ import PIL.ImageOps
 
 _CITATION = """\
 @InProceedings{huggingface:dataset,
-title = {body-measurements-dataset},
+title = {facial-emotion-recognition-dataset},
 author = {TrainingDataPro},
 year = {2023}
 }
 """
 
 _DESCRIPTION = """\
-The dataset consists of a compilation of people's photos along with their
-corresponding body measurements. It is designed to provide information and
-insights into the physical appearances and body characteristics of individuals.
-The dataset includes a diverse range of subjects representing different age
-groups, genders, and ethnicities. 
-
-The photos are captured in a standardized manner, depicting individuals in a
-front and side positions.
-The images aim to capture the subjects' physical appearance using appropriate
-lighting and angles that showcase their body proportions accurately.
-
-The dataset serves various purposes, including:
-- research projects
-- body measurement analysis
-- fashion or apparel industry applications
-- fitness and wellness studies
-- anthropometric studies for ergonomic design in various fields
+The dataset consists of images capturing people displaying 7 distinct emotions
+(anger, contempt, disgust, fear, happiness, sadness and surprise).
+Each image in the dataset represents one of these specific emotions,
+enabling researchers and machine learning practitioners to study and develop
+models for emotion recognition and analysis.
+The images encompass a diverse range of individuals, including different
+genders, ethnicities, and age groups*. The dataset aims to provide
+a comprehensive representation of human emotions, allowing for a wide range of
+use cases.
 """
-_NAME = 'body-measurements-dataset'
+_NAME = 'facial-emotion-recognition-dataset'
 
 _HOMEPAGE = f"https://huggingface.co/datasets/TrainingDataPro/{_NAME}"
 
@@ -43,141 +35,73 @@ _LICENSE = "cc-by-nc-nd-4.0"
 _DATA = f"https://huggingface.co/datasets/TrainingDataPro/{_NAME}/resolve/main/data/"
 
 
-class BodyMeasurementsDataset(datasets.GeneratorBasedBuilder):
+class FacialEmotionRecognitionDataset(datasets.GeneratorBasedBuilder):
 
     def _info(self):
-        return datasets.DatasetInfo(
-            description=_DESCRIPTION,
-            features=datasets.Features({
-                'front_img': datasets.Image(),
-                'selfie_img': datasets.Image(),
-                'side_img': datasets.Image(),
-                "arm_circumference_cm": datasets.Value('string'),
-                "arm_length_cm": datasets.Value('string'),
-                "back_build_cm": datasets.Value('string'),
-                "calf_circumference_cm": datasets.Value('string'),
-                "chest_circumference_cm": datasets.Value('string'),
-                "crotch_height_cm": datasets.Value('string'),
-                "front_build_cm": datasets.Value('string'),
-                "hips_circumference_cm": datasets.Value('string'),
-                "leg_length_cm": datasets.Value('string'),
-                "neck_circumference_cm": datasets.Value('string'),
-                "neck_pelvis_length_front_cm": datasets.Value('string'),
-                "neck_waist_length_back_cm": datasets.Value('string'),
-                "neck_waist_length_front_cm": datasets.Value('string'),
-                "pelvis_circumference_cm": datasets.Value('string'),
-                "shoulder_length_cm": datasets.Value('string'),
-                "shoulder_width_cm": datasets.Value('string'),
-                "thigh_circumference_cm": datasets.Value('string'),
-                "under_chest_circumference_cm": datasets.Value('string'),
-                "upper_arm_length_cm": datasets.Value('string'),
-                "waist_circumference_cm": datasets.Value('string'),
-                "height": datasets.Value('string'),
-                "weight": datasets.Value('string'),
-                "age": datasets.Value('string'),
-                "gender": datasets.Value('string'),
-                "race": datasets.Value('string'),
-                "profession": datasets.Value('string'),
-                "arm_circumference": datasets.Image(),
-                "arm_length": datasets.Image(),
-                "back_build": datasets.Image(),
-                "calf_circumference": datasets.Image(),
-                "chest_circumference": datasets.Image(),
-                "crotch_height": datasets.Image(),
-                "front_build": datasets.Image(),
-                "hips_circumference": datasets.Image(),
-                "leg_length": datasets.Image(),
-                "neck_circumference": datasets.Image(),
-                "neck_pelvis_length_front": datasets.Image(),
-                "neck_waist_length_back": datasets.Image(),
-                "neck_waist_length_front": datasets.Image(),
-                "pelvis_circumference": datasets.Image(),
-                "shoulder_length": datasets.Image(),
-                "shoulder_width": datasets.Image(),
-                "thigh_circumference": datasets.Image(),
-                "under_chest_circumference": datasets.Image(),
-                "upper_arm_length": datasets.Image(),
-                "waist_circumference": datasets.Image()
-            }),
-            supervised_keys=None,
-            homepage=_HOMEPAGE,
-            citation=_CITATION,
-            license=_LICENSE)
+        return datasets.DatasetInfo(description=_DESCRIPTION,
+                                    features=datasets.Features({
+                                        'set_id': datasets.Value('int32'),
+                                        'neutral': datasets.Image(),
+                                        'anger': datasets.Image(),
+                                        'contempt': datasets.Image(),
+                                        'disgust': datasets.Image(),
+                                        "fear": datasets.Image(),
+                                        "happy": datasets.Image(),
+                                        "sad": datasets.Image(),
+                                        "surprised": datasets.Image(),
+                                        "age": datasets.Value('int8'),
+                                        "gender": datasets.Value('string'),
+                                        "country": datasets.Value('string')
+                                    }),
+                                    supervised_keys=None,
+                                    homepage=_HOMEPAGE,
+                                    citation=_CITATION,
+                                    license=_LICENSE)
 
     def _split_generators(self, dl_manager):
-        files = dl_manager.download_and_extract(f"{_DATA}files.zip")
-        proofs = dl_manager.download_and_extract(f"{_DATA}proofs.zip")
+        images = dl_manager.download_and_extract(f"{_DATA}images.zip")
         annotations = dl_manager.download(f"{_DATA}{_NAME}.csv")
-        files = dl_manager.iter_files(files)
-        proofs = dl_manager.iter_files(proofs)
+        images = dl_manager.iter_files(images)
         return [
             datasets.SplitGenerator(name=datasets.Split.TRAIN,
                                     gen_kwargs={
-                                        "files": files,
-                                        'proofs': proofs,
+                                        "images": images,
                                         'annotations': annotations
                                     }),
         ]
 
-    def _generate_examples(self, files, proofs, annotations):
-        files = list(files)
-        files = [files[i:i + 4] for i in range(0, len(files), 4)]
-        proofs = list(proofs)
-        proofs = [proofs[i:i + 20] for i in range(0, len(proofs), 20)]
-
-        for idx, (files_dir, proofs_dir) in enumerate(zip(files, proofs)):
-            data = {}
-            for file in files_dir:
-                if 'front_img' in file:
-                    data['front_img'] = file
-                elif 'selfie_img' in file:
-                    data['selfie_img'] = file
-                elif 'side_img' in file:
-                    data['side_img'] = file
-                elif 'measurements' in file:
-                    with open(file) as f:
-                        data.update(json.load(f))
-
-            for proof in proofs_dir:
-                if "arm_circumference" in proof:
-                    data['arm_circumference'] = proof
-                elif 'upper_arm_length' in proof:
-                    data['upper_arm_length'] = proof
-                elif 'arm_length' in proof:
-                    data['arm_length'] = proof
-                elif 'back_build' in proof:
-                    data['back_build'] = proof
-                elif 'calf_circumference' in proof:
-                    data['calf_circumference'] = proof
-                elif 'under_chest_circumference' in proof:
-                    data['under_chest_circumference'] = proof
-                elif 'chest_circumference' in proof:
-                    data['chest_circumference'] = proof
-                elif 'crotch_height' in proof:
-                    data['crotch_height'] = proof
-                elif 'front_build' in proof:
-                    data['front_build'] = proof
-                elif 'hips_circumference' in proof:
-                    data['hips_circumference'] = proof
-                elif 'leg_length' in proof:
-                    data['leg_length'] = proof
-                elif 'neck_circumference' in proof:
-                    data['neck_circumference'] = proof
-                elif 'neck_pelvis_length_front' in proof:
-                    data['neck_pelvis_length_front'] = proof
-                elif 'neck_waist_length_back' in proof:
-                    data['neck_waist_length_back'] = proof
-                elif 'neck_waist_length_front' in proof:
-                    data['neck_waist_length_front'] = proof
-                elif 'pelvis_circumference' in proof:
-                    data['pelvis_circumference'] = proof
-                elif 'shoulder_length' in proof:
-                    data['shoulder_length'] = proof
-                elif 'shoulder_width' in proof:
-                    data['shoulder_width'] = proof
-                elif 'thigh_circumference' in proof:
-                    data['thigh_circumference'] = proof
-                elif 'waist_circumference' in proof:
-                    data['waist_circumference'] = proof
+    def _generate_examples(self, images, annotations):
+        annotations_df = pd.read_csv(annotations, sep=';')
+        images = list(images)
+        images = [sorted(images)[i:i + 8] for i in range(0, len(images), 8)]
+
+        for idx, images_set in enumerate(images):
+            set_id = images_set.split('/')[-2]
+            data = {'set_id': set_id}
+
+            for file in images_set:
+                if 'neutral' in file.lower():
+                    data['neutral'] = file
+                elif 'anger' in file.lower():
+                    data['anger'] = file
+                elif 'contempt' in file.lower():
+                    data['contempt'] = file
+                elif 'disgust' in file.lower():
+                    data['disgust'] = file
+                elif 'fear' in file.lower():
+                    data['fear'] = file
+                elif 'happy' in file.lower():
+                    data['happy'] = file
+                elif 'sad' in file.lower():
+                    data['sad'] = file
+                elif 'surprised' in file.lower():
+                    data['surprised'] = file
+
+            data['age'] = annotations_df.loc[annotations_df['set_id'] ==
+                                             set_id]['age'].values[0]
+            data['gender'] = annotations_df.loc[annotations_df['set_id'] ==
+                                                set_id]['gender'].values[0]
+            data['country'] = annotations_df.loc[annotations_df['set_id'] ==
+                                                 set_id]['country'].values[0]
 
             yield idx, data