update

Dockerfile

@@ -8,7 +8,7 @@ RUN python3 -m pip install --no-cache-dir --upgrade -r /code/requirements.txt
 
 COPY . .
 
-CMD ["panel", "serve", "/code/app.ipynb", "--address", "0.0.0.0", "--port", "7860",  "--allow-websocket-origin", "*"]
+CMD ["panel", "serve", "/code/app.py", "--address", "0.0.0.0", "--port", "7860",  "--allow-websocket-origin", "*"]
 
 RUN mkdir /.cache
 RUN chmod 777 /.cache

app.ipynb

@@ -1,173 +0,0 @@
-{
- "cells": [
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "8cd1e865-53d5-460b-8bae-5658e3aa3d16",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "import panel as pn\n",
-    "pn.extension()\n",
-    "import requests\n",
-    "import random\n",
-    "import PIL\n",
-    "from PIL import Image\n",
-    "import io\n",
-    "from transformers import CLIPProcessor, CLIPModel\n",
-    "import numpy as np"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "e8570053-0b83-421b-95c2-695b6c709ba1",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "pn.extension('texteditor', template=\"bootstrap\", sizing_mode='stretch_width')\n",
-    "\n",
-    "pn.state.template.param.update(\n",
-    "    main_max_width=\"690px\",\n",
-    "    header_background=\"#F08080\",\n",
-    ")"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "ca65cc07-8181-4259-8770-9c780621eb78",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "# File input widget\n",
-    "file_input = pn.widgets.FileInput()\n",
-    "\n",
-    "# Button widget\n",
-    "compute_button = pn.widgets.Button(name=\"Compute\")\n",
-    "\n",
-    "# Text input widget\n",
-    "text_input = pn.widgets.TextInput(name='Possible class names (e.g., cat, dog)', placeholder='cat, dog')"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "f3691594-df8c-4d03-99e8-db4d3b2520c0",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "def normalize_image(value, width=600):\n",
-    "    \"\"\"\n",
-    "    normalize image to RBG channels and to the same size\n",
-    "    \"\"\"\n",
-    "    if value: \n",
-    "        b = io.BytesIO(value)\n",
-    "        image = PIL.Image.open(b).convert(\"RGB\")\n",
-    "    else: \n",
-    "        url = \"http://images.cocodataset.org/val2017/000000039769.jpg\"\n",
-    "        image = Image.open(requests.get(url, stream=True).raw)\n",
-    "    aspect = image.size[1] / image.size[0]\n",
-    "    height = int(aspect * width)\n",
-    "    return image.resize((width, height), PIL.Image.LANCZOS)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "5b139802-c9d6-4493-acb2-5051343c1ecc",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "def image_classification(image):\n",
-    "    model = CLIPModel.from_pretrained(\"openai/clip-vit-large-patch14\")\n",
-    "    processor = CLIPProcessor.from_pretrained(\"openai/clip-vit-large-patch14\")\n",
-    "    possible_categories = text_input.value.split(\",\")\n",
-    "    if text_input.value == '':\n",
-    "        possible_categories = ['cat', ' dog']\n",
-    "    inputs = processor(text=possible_categories, images=image, return_tensors=\"pt\", padding=True)\n",
-    "    \n",
-    "    outputs = model(**inputs)\n",
-    "    logits_per_image = outputs.logits_per_image # this is the image-text similarity score\n",
-    "    probs = logits_per_image.softmax(dim=1)\n",
-    "    return probs.detach().numpy()"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "6b6f0ce5-03a5-4a14-b0b7-74c8190ce928",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "def get_result(_):\n",
-    "    image = normalize_image(file_input.value)\n",
-    "\n",
-    "    result = image_classification(image)\n",
-    "    \n",
-    "    possible_categories = text_input.value.split(\",\")\n",
-    "    if text_input.value == '':\n",
-    "        possible_categories = ['cat', ' dog']\n",
-    "\n",
-    "    progress_bars = pn.Column(*[\n",
-    "        pn.Row(\n",
-    "            possible_categories[i], \n",
-    "            pn.indicators.Progress(name='', value=int(j*100), width=500))\n",
-    "        for i, j in enumerate(result[0])\n",
-    "    ])\n",
-    "    return progress_bars\n",
-    "    "
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "6fd5a63f-012a-419c-8386-22b5b8ff243f",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "# Bind the get_image function with the button widget\n",
-    "interactive_result = pn.bind(get_result, compute_button)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "399189f1-4ff6-4f4b-b050-76e9a46443dd",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "# layout\n",
-    "pn.Column(\n",
-    "    \"## \\U0001F60A Upload an image file and start classifying!\",\n",
-    "    file_input,\n",
-    "    pn.bind(pn.panel, file_input),\n",
-    "    text_input, \n",
-    "    compute_button,\n",
-    "    interactive_result\n",
-    ").servable(title=\"Panel Image Classification Demo\")"
-   ]
-  }
- ],
- "metadata": {
-  "kernelspec": {
-   "display_name": "Python 3 (ipykernel)",
-   "language": "python",
-   "name": "python3"
-  },
-  "language_info": {
-   "codemirror_mode": {
-    "name": "ipython",
-    "version": 3
-   },
-   "file_extension": ".py",
-   "mimetype": "text/x-python",
-   "name": "python",
-   "nbconvert_exporter": "python",
-   "pygments_lexer": "ipython3",
-   "version": "3.10.11"
-  }
- },
- "nbformat": 4,
- "nbformat_minor": 5
-}

app.py

@@ -0,0 +1,122 @@
+import random
+import panel as pn
+import requests
+from PIL import Image
+
+from transformers import CLIPProcessor, CLIPModel
+from typing import List, Tuple
+
+pn.extension("texteditor", sizing_mode="stretch_width")
+
+
+def set_random_url(_):
+    if random.randint(0, 1) == 0:
+        api_url = "https://api.thecatapi.com/v1/images/search"
+    else:
+        api_url = "https://api.thedogapi.com/v1/images/search"
+    with requests.get(api_url) as resp:
+        resp.raise_for_status()
+        url = resp.json()[0]["url"]
+    image_url.value = url
+
+
+@pn.cache
+def load_processor_model(
+    processor_name: str, model_name: str
+) -> Tuple[CLIPProcessor, CLIPModel]:
+    processor = CLIPProcessor.from_pretrained(processor_name)
+    model = CLIPModel.from_pretrained(model_name)
+    return processor, model
+
+
+@pn.cache
+def open_image_url(image_url: str) -> Image:
+    with requests.get(image_url, stream=True) as resp:
+        resp.raise_for_status()
+        image = Image.open(resp.raw)
+    return image
+
+
+def get_similarity_scores(class_items: List[str], image: Image) -> List[float]:
+    processor, model = load_processor_model(
+        "openai/clip-vit-base-patch32", "openai/clip-vit-base-patch32"
+    )
+    inputs = processor(
+        text=class_items,
+        images=[image],
+        return_tensors="pt",  # pytorch tensors
+    )
+    outputs = model(**inputs)
+    logits_per_image = outputs.logits_per_image
+    class_likelihoods = logits_per_image.softmax(dim=1).detach().numpy()
+    return class_likelihoods[0]
+
+
+def process_inputs(class_names: List[str], image_url: str):
+    """
+    High level function that takes in the user inputs and returns the
+    classification results as panel objects.
+    """
+    image = open_image_url(image_url)
+    class_items = class_names.split(",")
+    class_likelihoods = get_similarity_scores(class_items, image)
+
+    # build the results column
+    results_column = pn.Column("## 🎉 Here are the results!")
+
+    results_column.append(
+        pn.pane.Image(image, max_width=698, sizing_mode="scale_width")
+    )
+
+    for class_item, class_likelihood in zip(class_items, class_likelihoods):
+        row_label = pn.widgets.StaticText(
+            name=class_item.strip(), value=f"{class_likelihood:.2%}", margin=(0, 10)
+        )
+        row_bar = pn.indicators.Progress(
+            max=100,
+            value=int(class_likelihood * 100),
+            sizing_mode="stretch_width",
+            bar_color="secondary",
+            margin=(0, 10),
+        )
+        row_column = pn.Column(row_label, row_bar)
+        results_column.append(row_column)
+    return results_column
+
+# create widgets
+randomize_url = pn.widgets.Button(name="Randomize URL", align="end")
+
+image_url = pn.widgets.TextInput(
+    name="Image URL to classify",
+    value="https://cdn2.thecatapi.com/images/cct.jpg",
+)
+class_names = pn.widgets.TextInput(
+    name="Comma separated class names",
+    placeholder="Enter possible class names, e.g. cat, dog",
+    value="cat, dog, parrot",
+)
+
+input_widgets = pn.Column(
+    "## 😊 Click randomize or paste a URL to start classifying!",
+    pn.Row(image_url, randomize_url),
+    class_names,
+)
+
+# add interactivity
+randomize_url.on_click(set_random_url)
+interactive_result = pn.bind(
+    process_inputs, image_url=image_url, class_names=class_names
+)
+
+# create dashboard
+main = pn.WidgetBox(
+    input_widgets,
+    interactive_result,
+)
+
+pn.template.BootstrapTemplate(
+    title="Panel Image Classification Demo",
+    main=main,
+    main_max_width="min(50%, 698px)",
+    header_background="#F08080",
+).servable(title="Panel Image Classification Demo")