Update app.py

app.py

@@ -1,4 +1,3 @@
-# app.py
 from fastapi import FastAPI, File, UploadFile
 from transformers import pipeline
 from PIL import Image
@@ -6,39 +5,33 @@ import io
 
 app = FastAPI()
 
-# Load the model pipeline only once when the Space starts.
-# This makes it ready for all incoming requests.
+# Load the lightweight zero‑shot image classification model
+# `device=-1` forces the model to run on the CPU
 classifier = pipeline("zero-shot-image-classification",
-                      model="google/vit-base-patch16-224",
-                      device=-1)  # -1 forces the model to use the CPU
+                      model="sachin/tiny_clip",
+                      device=-1)
 
 @app.get("/")
-def root():
+def read_root():
     return {"message": "Fish Detector API is running. Use POST /detect"}
 
 @app.post("/detect")
 async def detect(file: UploadFile = File(...)):
-    # 1. Read and open the image from the uploaded file
+    # 1. Read and open the uploaded image
     contents = await file.read()
     image = Image.open(io.BytesIO(contents)).convert("RGB")
 
-    # 2. Define the labels for zero-shot classification.
-    #    These are the categories our model will choose from.
+    # 2. Define the two possible labels for the model to choose from
     candidate_labels = [
-        "a photo of a fish, with fins and scales",
-        "a photo of a person, an animal, or an object that is not a fish",
-        "an empty photo with no fish in it"
+        "a photo of a fish",
+        "a photo that does not contain a fish"
     ]
 
-    # 3. Run the image through the pipeline to get predictions
+    # 3. Run the zero‑shot classifier
     predictions = classifier(image, candidate_labels=candidate_labels)
 
-    # 4. The top prediction is the one with the highest score.
-    #    We check if its label is the fish-related one.
-    is_fish = predictions[0]['label'] == candidate_labels[0]
-
-    # 5. Return the result as a plain text string
-    if is_fish:
+    # 4. The top prediction is the one with the highest score
+    if predictions[0]['label'] == "a photo of a fish":
         return "fish"
     else:
         return "not a fish"