Delete handler.py

handler.py

@@ -1,82 +0,0 @@
-import os
-import json
-import torch
-import torch.nn as nn
-from PIL import Image
-import io
-from transformers import AutoImageProcessor, AutoModel, AutoModelForImageClassification
-
-class EndpointHandler:
-    def __init__(self, path=""):
-        self.device = "cuda" if torch.cuda.is_available() else "cpu"
-        
-        # قراءة معلومات الأصناف
-        meta_file = os.path.join(path, "meta.json")
-        with open(meta_file, "r") as f:
-            meta = json.load(f)
-        self.classes = meta["classes"]
-        self.id2label = {int(k): v for k, v in meta["id2label"].items()}
-        self.label2id = meta["label2id"]
-
-        # تحميل المعالج
-        self.processor = AutoImageProcessor.from_pretrained(path)
-
-        # تحميل الخلفية (Backbone)
-        backbone_path = os.path.join(path, "backbone")
-        self.backbone = AutoModel.from_pretrained(backbone_path).to(self.device)
-        self.backbone.eval()
-
-        # بناء هيكل النموذج الموحد
-        self.model = AutoModelForImageClassification.from_pretrained(
-            "microsoft/rad-dino",
-            num_labels=len(self.classes),
-            id2label=self.id2label,
-            label2id=self.label2id
-        )
-        
-        # ربط الخلفية
-        if hasattr(self.model, 'dinov2'):
-            self.model.dinov2 = self.backbone
-        elif hasattr(self.model, 'backbone'):
-            self.model.backbone = self.backbone
-
-        # تحميل أوزان رأس التصنيف
-        head_file = os.path.join(path, "classifier_head.pt")
-        state_dict = torch.load(head_file, map_location=self.device)
-        self.model.classifier.weight.copy_(state_dict["1.weight"])
-        self.model.classifier.bias.copy_(state_dict["1.bias"])
-        
-        self.model.to(self.device)
-        self.model.eval()
-
-    def __call__(self, data):
-        inputs = data.get("inputs", data)
-        
-        # معالجة الصورة القادمة أياً كانت صيغتها
-        if isinstance(inputs, str):
-            image = Image.open(inputs).convert("RGB")
-        elif isinstance(inputs, Image.Image):
-            image = inputs.convert("RGB")
-        elif isinstance(inputs, bytes):
-            image = Image.open(io.BytesIO(inputs)).convert("RGB")
-        else:
-            # إذا أرسلت كملف بايتس عبر الـ API
-            image = Image.open(io.BytesIO(inputs)).convert("RGB")
-
-        processed_inputs = self.processor(images=image, return_tensors="pt").to(self.device)
-        
-        with torch.no_grad():
-            outputs = self.model(**processed_inputs)
-            logits = outputs.logits
-            probs = torch.softmax(logits, dim=-1)[0]
-
-        results = []
-        for idx, prob in enumerate(probs):
-            results.append({
-                "label": self.id2label[idx],
-                "score": float(prob.item())
-            })
-        
-        # ترتيب النتائج من الأعلى للأقل
-        results = sorted(results, key=lambda x: x["score"], reverse=True)
-        return results