Project Implementation

app.py

@@ -0,0 +1,24 @@
+from utils.warnings import suppress_warnings
+from core.model_loader import load_model
+from ui.theme import OrangeRedTheme
+from ui.styles import CSS_STYLE
+from ui.layout import build_ui
+
+def main():
+    suppress_warnings()
+
+    model, processor = load_model()
+    theme = OrangeRedTheme()
+
+    demo = build_ui(model, processor)
+    demo.queue().launch(
+        theme=theme,
+        css=CSS_STYLE,
+        show_error=True,
+        server_name="0.0.0.0",
+        server_port=7860,
+        debug=True
+    )
+
+if __name__ == "__main__":
+    main()

core/model_loader.py

@@ -0,0 +1,47 @@
+import torch
+from transformers import BlipProcessor, BlipForConditionalGeneration
+
+MODEL_ID = "Salesforce/blip-image-captioning-base"
+DEVICE = torch.device("cpu")
+
+# Prompt templates
+PROMPTS = {
+    "Short Caption": "a photo of",
+    "Detailed Caption": "a detailed description of",
+    "Creative Caption": "an artistic depiction of",
+    "Image Explanation": "explain what is happening in this image"
+}
+
+def load_model():
+    processor = BlipProcessor.from_pretrained(MODEL_ID)
+    model = BlipForConditionalGeneration.from_pretrained(MODEL_ID)
+    model.to(DEVICE)
+    model.eval()
+    return model, processor
+
+def generate_caption(
+    model,
+    processor,
+    image,
+    style
+):
+    prompt = PROMPTS.get(style, "a photo of")
+
+    inputs = processor(
+        images=image,
+        text=prompt,
+        return_tensors="pt"
+    ).to(DEVICE)
+
+    max_len = 30 if style == "Short Caption" else 60
+
+    with torch.inference_mode():
+        output_ids = model.generate(
+            **inputs,
+            max_length=max_len
+        )
+
+    return processor.decode(
+        output_ids[0],
+        skip_special_tokens=True
+    )

requirements.txt

@@ -0,0 +1,4 @@
+gradio>=4.0.0
+torch>=2.0.0
+transformers>=4.35.0
+pillow

ui/layout.py

@@ -0,0 +1,53 @@
+import gradio as gr
+from core.model_loader import generate_caption
+
+def build_ui(model, processor):
+
+    def run_caption(image, style):
+        if image is None:
+            return "Please upload an image."
+        image = image.convert("RGB")
+        return generate_caption(
+            model=model,
+            processor=processor,
+            image=image,
+            style=style
+        )
+
+    with gr.Blocks() as demo:
+        gr.Markdown("## 🖼️ BLIP Image Captioning (Zero-GPU)")
+        gr.Markdown(
+            "Generate captions or explanations from images using a CPU-only BLIP model."
+        )
+
+        with gr.Row():
+            image_input = gr.Image(
+                type="pil",
+                label="Upload Image"
+            )
+
+            with gr.Column():
+                style_select = gr.Dropdown(
+                    choices=[
+                        "Short Caption",
+                        "Detailed Caption",
+                        "Creative Caption",
+                        "Image Explanation"
+                    ],
+                    value="Detailed Caption",
+                    label="Caption Style"
+                )
+                output_text = gr.Textbox(
+                    label="Generated Output",
+                    lines=4
+                )
+
+        generate_btn = gr.Button("Generate")
+
+        generate_btn.click(
+            fn=run_caption,
+            inputs=[image_input, style_select],
+            outputs=output_text
+        )
+
+    return demo

ui/styles.py

@@ -0,0 +1,16 @@
+CSS_STYLE = """
+#container {
+    max-width: 1280px;
+    margin: auto;
+}
+
+@media (min-width: 1600px) {
+    #container {
+        max-width: 1440px;
+    }
+}
+
+#title h1 {
+    font-size: 2.4em !important;
+}
+"""

ui/theme.py

@@ -0,0 +1,28 @@
+from gradio.themes import Soft
+from gradio.themes.utils import colors, fonts, sizes
+
+colors.orange_red = colors.Color(
+    name="orange_red",
+    c50="#FFF0E5", c100="#FFE0CC", c200="#FFC299", c300="#FFA366",
+    c400="#FF8533", c500="#FF4500", c600="#E63E00", c700="#CC3700",
+    c800="#B33000", c900="#992900", c950="#802200",
+)
+
+class OrangeRedTheme(Soft):
+    def __init__(self):
+        super().__init__(
+            primary_hue=colors.orange_red,
+            secondary_hue=colors.orange_red,
+            neutral_hue=colors.slate,
+            text_size=sizes.text_lg,
+            font=(fonts.GoogleFont("Outfit"), "Arial", "sans-serif"),
+            font_mono=(fonts.GoogleFont("IBM Plex Mono"), "monospace"),
+        )
+        super().set(
+            body_background_fill="linear-gradient(135deg, *primary_200, *primary_100)",
+            button_primary_background_fill="linear-gradient(90deg, *secondary_500, *secondary_600)",
+            button_primary_background_fill_hover="linear-gradient(90deg, *secondary_600, *secondary_700)",
+            button_primary_text_color="white",
+            block_border_width="3px",
+            block_shadow="*shadow_drop_lg",
+        )

utils/warnings.py

@@ -0,0 +1,4 @@
+import warnings
+
+def suppress_warnings():
+    warnings.filterwarnings("ignore")