Update app.py

app.py

@@ -1,26 +1,24 @@
 import gradio as gr
-from dotenv import load_dotenv
-from roboflow import Roboflow
 import tempfile
 import os
-import requests
-import numpy as np  # Import numpy to handle image slices
-from sahi.predict import get_sliced_prediction  # SAHI slicing inference
-import supervision as sv  # For annotating images with results
-
-# Muat variabel lingkungan dari file .env
-load_dotenv()
-api_key = os.getenv("ROBOFLOW_API_KEY")
-workspace = os.getenv("ROBOFLOW_WORKSPACE")
-project_name = os.getenv("ROBOFLOW_PROJECT")
-model_version = int(os.getenv("ROBOFLOW_MODEL_VERSION"))
-
-# Inisialisasi Roboflow menggunakan data yang diambil dari secrets
-rf = Roboflow(api_key=api_key)
-project = rf.workspace(workspace).project(project_name)
-model = project.version(model_version).model
-
-# Fungsi untuk menangani input dan output gambar
+from sahi import AutoDetectionModel
+from sahi.predict import get_sliced_prediction
+from PIL import Image
+
+# Inisialisasi model deteksi menggunakan SAHI
+model_path = "best.pt"  # Ganti dengan path model YOLO lokal Anda
+confidence_threshold = 0.6  # Threshold kepercayaan
+sahi_device = 'cuda'  # Ganti dengan 'cpu' jika tidak menggunakan GPU
+
+# Memuat model YOLO menggunakan SAHI
+sahi_model = AutoDetectionModel.from_pretrained(
+    model_type="yolov11",  # Tipe model YOLO, sesuaikan jika model YOLO yang digunakan berbeda
+    model_path=model_path,
+    confidence_threshold=confidence_threshold,
+    device=sahi_device
+)
+
+# Fungsi untuk deteksi objek menggunakan SAHI
 def detect_objects(image):
     # Simpan gambar yang diupload sebagai file sementara
     with tempfile.NamedTemporaryFile(delete=False, suffix=".jpg") as temp_file:
@@ -28,85 +26,64 @@ def detect_objects(image):
         temp_file_path = temp_file.name
 
     try:
-        # Perform sliced inference with SAHI using InferenceSlicer
-        def callback(image_slice: np.ndarray) -> sv.Detections:
-            results = model.infer(image_slice)[0]  # Perform inference on each slice
-            return sv.Detections.from_inference(results)
-
-        # Configure the SAHI Slicer with specific slice dimensions and overlap
-        slicer = sv.InferenceSlicer(
-            callback=callback,
-            slice_wh=(320, 320),  # Adjust slice dimensions as needed
-            overlap_wh=(64, 64),  # Adjust overlap in pixels (DO NOT use overlap_ratio_wh here)
-            overlap_filter=sv.OverlapFilter.NON_MAX_SUPPRESSION,  # Filter overlapping detections
-            iou_threshold=0.5,  # Intersection over Union threshold for NMS
+        # Lakukan prediksi pada gambar menggunakan SAHI
+        results = get_sliced_prediction(
+            image=image,
+            detection_model=sahi_model,
+            slice_height=512,  # Ukuran potongan gambar (bisa disesuaikan)
+            slice_width=512,
+            overlap_height_ratio=0.2,
+            overlap_width_ratio=0.2
         )
 
-        # Run slicing-based inference
-        detections = slicer(image)
-
-        # Annotate the results on the image
-        box_annotator = sv.BoxAnnotator()
-        label_annotator = sv.LabelAnnotator()
-
-        annotated_image = box_annotator.annotate(
-            scene=image.copy(), detections=detections)
-        
-        annotated_image = label_annotator.annotate(
-            scene=annotated_image, detections=detections)
-
-        # Save the annotated image
-        output_image_path = "/tmp/prediction_visual.png"
-        annotated_image.save(output_image_path)
-
-        # Count the number of detected objects per class
+        # Menghitung jumlah objek per kelas
         class_count = {}
-        total_count = 0
+        total_count = 0  # Menyimpan total jumlah objek
 
-        for prediction in detections:
-            class_name = prediction.class_id  # or prediction.class_name if available
+        for prediction in results.object_prediction_list:
+            class_name = prediction.category.name  # Nama kelas objek
             class_count[class_name] = class_count.get(class_name, 0) + 1
-            total_count += 1  # Increment the total object count
+            total_count += 1  # Menambah jumlah objek
 
-        # Create a result text with object counts
+        # Menyusun output berupa string hasil perhitungan
         result_text = "Detected Objects:\n\n"
         for class_name, count in class_count.items():
             result_text += f"{class_name}: {count}\n"
-        result_text += f"\nTotal objects detected: {total_count}"
+        result_text += f"\nTotal Objects: {total_count}"
+
+        # Menyimpan gambar dengan prediksi
+        output_image_path = "/tmp/prediction.jpg"
+        results.save(output_image_path)  # Menyimpan gambar dengan prediksi
 
-    except requests.exceptions.HTTPError as http_err:
-        # Handle HTTP errors
-        result_text = f"HTTP error occurred: {http_err}"
-        output_image_path = temp_file_path  # Return the original image in case of error
     except Exception as err:
-        # Handle other errors
+        # Menangani kesalahan lain
         result_text = f"An error occurred: {err}"
-        output_image_path = temp_file_path  # Return the original image in case of error
+        output_image_path = temp_file_path  # Kembalikan gambar asli jika terjadi error
 
-    # Clean up temporary files
+    # Hapus file sementara setelah prediksi
     os.remove(temp_file_path)
-    
+
     return output_image_path, result_text
 
-# Create the Gradio interface
+# Membuat antarmuka Gradio dengan tata letak fleksibel
 with gr.Blocks() as iface:
     with gr.Row():
         with gr.Column():
             input_image = gr.Image(type="pil", label="Input Image")
         with gr.Column():
-            output_image = gr.Image(label="Detected Objects")
+            output_image = gr.Image(label="Detect Object")
         with gr.Column():
-            output_text = gr.Textbox(label="Object Count")
-    
-    # Button to trigger object detection
-    detect_button = gr.Button("Detect Objects")
-    
-    # Link the button to the detect_objects function
+            output_text = gr.Textbox(label="Counting Object")
+
+    # Tombol untuk memproses input
+    detect_button = gr.Button("Detect")
+
+    # Hubungkan tombol dengan fungsi deteksi
     detect_button.click(
         fn=detect_objects, 
         inputs=input_image, 
         outputs=[output_image, output_text]
     )
 
-# Launch the interface
+# Menjalankan antarmuka
 iface.launch()