working on unsupervised annnotation

2_data_splitting.py

@@ -5,14 +5,19 @@
 
 dataset_name = "photo_album"
 
+# This just takes a sample of the original dataset and creates
+# a new dataset which is has a random 300 samples.
+# For real work we probably would just want to tag this data. Something like training,
+# test, validation
+
 if __name__ == "__main__":
     print("starting")
     dataset = fo.load_dataset(dataset_name)
     # A DatasetView is a shallow copy of the original dataset.This means,
     # if your dataset is persisted, any data or schema changes to the view will be persisted to the original dataset.
     # If you don't want to affect the original dataset you must make the view a clone of the original data.
     shuffle_view = dataset.shuffle()
-    training_view = shuffle_view[0:300].clone()
+    training_view = shuffle_view[0:300].clone("play_photos", persistent=True)
     training_view.save()
 
     print("finished")

4_generate_labels.py

@@ -0,0 +1,27 @@
+from http.cookiejar import debug
+
+import fiftyone as fo
+import fiftyone.zoo as foz
+
+
+if __name__ == '__main__':
+    dataset = fo.load_dataset("play_photos")
+    clip = foz.load_zoo_model(
+        "clip-vit-base32-torch",
+        text_prompt="A photo of a",
+        classes= ["boy", "girl", "man", "woman", "people", "dog", "cat", "bird", "insect", "monkey", "crustacean",
+                  "fish", "animal", "plant", "flower", "landscape", "architecture", "not an animal, plant, landscape, person, or building"])
+    # alexnet = foz.load_zoo_model("alexnet-imagenet-torch")
+    # dense201 = foz.load_zoo_model("densenet201-imagenet-torch")
+    # fasterrcnn = foz.load_zoo_model("faster-rcnn-resnet50-fpn-coco-torch")
+    # yoloseg = foz.load_zoo_model("yolo11x-seg-coco-torch")
+
+    dataset.apply_model(clip, label_field="default_prediction")
+    dataset.set_values("ground_truth", fo.Classification(label=str(dataset.values("default_prediction.label"))))
+    # dataset.apply_model(dense201, label_field="dense201")
+    # dataset.apply_model(alexnet, label_field="alexnet")
+    # dataset.apply_model(fasterrcnn, label_field="faster_rcnn")
+    # dataset.apply_model(yoloseg, label_field="yolo_seg")
+
+    session = fo.launch_app(dataset)
+    session.wait()