dev -> master

README.md

@@ -1,5 +1,5 @@
 ---
-title: FindYourTwins
+title: FindYourSiblings
 emoji: 📉
 colorFrom: green
 colorTo: red

app.py

@@ -11,16 +11,12 @@ model = AutoModel.from_pretrained(model_ckpt)
 hidden_dim = model.config.hidden_size
 dataset = load_dataset("BounharAbdelaziz/Face-Aging-Dataset")
 
-TOP_K = 1000
-BASE_COUNT=4
-MAX_COUNT = 10
-
 with gr.Blocks() as demo:
     gr.Markdown(
         """
         # Find Your Twins
 
-        Upload your face and find the most similar faces from [Face Aging Dataset](https://huggingface.co/datasets/BounharAbdelaziz/Face-Aging-Dataset) using Google's [VIT](https://huggingface.co/google/vit-base-patch16-224-in21k) model. For best results please use 1x1 ratio face images, take a look at examples. Also increasing count in the advanced section results with more accurate searches. Disclaimer, this demo doesn't find your twins :), it finds similar face parts, shapes, features(nose, cheek, face, forehead shapes) that are encoded in the model. The Vector similarity search is powered by [Upstash Vector](https://upstash.com) 🚀. You can check our blog [post](https://huggingface.co/blog/omerXfaruq/serverless-image-similarity-with-upstash-vector) to learn more.
+        Upload your face and find the most similar people from [Face Aging Dataset](https://huggingface.co/datasets/BounharAbdelaziz/Face-Aging-Dataset) using Google's [VIT](https://huggingface.co/google/vit-base-patch16-224-in21k) model. The task of finding most similar vectors is powered by [Upstash Vector](https://upstash.com) 🚀. Check our blog post *here*.
         """
     )
 
@@ -28,27 +24,25 @@ with gr.Blocks() as demo:
         with gr.Row():
             with gr.Column(scale=1):
                 input_image = gr.Image(type="pil")
-            with gr.Column(scale=2):
-                output_images = gr.Gallery()
+            with gr.Column(scale=3):
+                output_image = gr.Gallery(height=800)
+
 
-        @input_image.change(inputs=input_image, outputs=output_images)
+        @input_image.change(inputs=input_image, outputs=output_image)
         async def find_similar_faces(image):
             if image is None:
                 return None
             inputs = extractor(images=image, return_tensors="pt")
             outputs = model(**inputs)
             embed = outputs.last_hidden_state[0][0]
-            result = await index.query(vector=embed.tolist(), top_k=TOP_K)
-            return [dataset["train"][int(vector.id)]["image"] for vector in result[:BASE_COUNT]]
+            result = await index.query(vector=embed.tolist(), top_k=4)
+            return [dataset["train"][int(vector.id)]["image"] for vector in result]
+
 
         gr.Examples(
-            examples=[
-                dataset["train"][6]["image"],
-                dataset["train"][7]["image"],
-                dataset["train"][8]["image"],
-            ],
+            examples=[dataset["train"][6]["image"]],
             inputs=input_image,
-            outputs=output_images,
+            outputs=output_image,
             fn=find_similar_faces,
             cache_examples=False,
         )
@@ -57,47 +51,21 @@ with gr.Blocks() as demo:
         with gr.Row():
             with gr.Column(scale=1):
                 adv_input_image = gr.Image(type="pil")
-                adv_image_count = gr.Slider(1, MAX_COUNT, BASE_COUNT, label="Image Count")
-                adv_button = gr.Button("Submit")
+                adv_image_count = gr.Number(9, label="Image Count")
 
-            with gr.Column(scale=2):
-                adv_output_images = gr.Gallery()
+            with gr.Column(scale=3):
+                adv_output_image = gr.Gallery(height=1000)
 
+
+        @adv_input_image.upload(inputs=[adv_input_image, adv_image_count], outputs=[adv_output_image])
         async def find_similar_faces(image, count):
-            if image is None:
-                return None
             inputs = extractor(images=image, return_tensors="pt")
             outputs = model(**inputs)
             embed = outputs.last_hidden_state[0][0]
             result = await index.query(
-                vector=embed.tolist(), top_k=TOP_K
+                vector=embed.tolist(), top_k=max(1, min(19, count))
             )
-            return [dataset["train"][int(vector.id)]["image"] for vector in result[:int(count)]]
-
-        adv_button.click(
-            fn=find_similar_faces,
-            inputs=[adv_input_image, adv_image_count],
-            outputs=[adv_output_images],
-        )
-        adv_input_image.change(
-            fn=find_similar_faces,
-            inputs=[adv_input_image, adv_image_count],
-            outputs=[adv_output_images],
-        )
-        gr.Examples(
-            examples=[
-                [dataset["train"][6]["image"], BASE_COUNT],
-                [dataset["train"][7]["image"], BASE_COUNT],
-                [dataset["train"][8]["image"], BASE_COUNT],
-            ],
-            inputs=[adv_input_image, adv_image_count],
-            outputs=adv_output_images,
-            fn=find_similar_faces,
-            cache_examples=False,
-        )
-
-
+            return [dataset["train"][int(vector.id)]["image"] for vector in result]
 
 if __name__ == "__main__":
-    demo.queue(default_concurrency_limit=40)
-    demo.launch()
+    demo.launch(debug=True)

requirements.txt

@@ -1,4 +1,4 @@
-torch
+torchvision
 transformers
 datasets
 upstash-vector