Upload 4 files

app.py

@@ -1,3 +1,4 @@
+
 import gradio as gr
 import re
 import os
@@ -7,16 +8,92 @@ import torch
 from PIL import Image
 from pathlib import Path
 
+from models.experimental import attempt_load
+from utils.datasets import LoadImage
+from utils.general import check_img_size, non_max_suppression, scale_coords, set_logging
+from utils.torch_utils import select_device
+import cv2
+#sudo apt-get install poppler-utils Necesario
+
+key = str(os.environ.get('key'))
 
 def check_image(image):
     try:
         images = convert_from_path(Path(image.name), fmt="jpeg", size=(960,1280))
-        return images[0]
+        return images
     except:
-        return Image.open(image)
+        return [Image.open(image)]
+
+def crop(files = '', #files
+           weights = 'yolov7.pt', #model.pt path(s)
+           classes = None, #filter by class: --class 0, or --class 0 2 3
+           imgsz = 640, #inference size (pixels)
+           device = '', #cuda device, i.e. 0 or 0,1,2,3 or cpu
+           conf_thres = 0.25, #object confidence threshold
+           iou_thres = 0.45, #IOU threshold for NMS
+           augment = False, #augmented inference
+           agnostic_nms = False): #class-agnostic NMS 
+
+    # Initialize
+    set_logging()
+    device = select_device(device)
+    half = device.type != 'cpu'  # half precision only supported on CUDA
+
+    # Load model
+    model = attempt_load(weights, map_location=device)  # load FP32 model
+    stride = int(model.stride.max())  # model stride
+    imgsz = check_img_size(imgsz, s=stride)  # check img_size
+
+    if half:
+        model.half()  # to FP16
+
+    # Set Dataloader
+    dataset = LoadImage(files = files, img_size=imgsz, stride=stride)
+
+    # Get names and colors
+    names = model.module.names if hasattr(model, 'module') else model.names
+    
+    # Run inference
+    if device.type != 'cpu':
+        model(torch.zeros(1, 3, imgsz, imgsz).to(device).type_as(next(model.parameters())))  # run once
+    old_img_w = old_img_h = imgsz
+    old_img_b = 1
+    list_cropobj = []
+    for img, img0s in dataset:
+        img = torch.from_numpy(img).to(device)
+        img = img.half() if half else img.float()  # uint8 to fp16/32
+        img /= 255.0  # 0 - 255 to 0.0 - 1.0
+        if img.ndimension() == 3:
+            img = img.unsqueeze(0)
+
+        # Inference
+        with torch.no_grad():   # Calculating gradients would cause a GPU memory leak
+            pred = model(img, augment=augment)[0]
+
+        # Apply NMS
+        pred = non_max_suppression(pred, conf_thres, iou_thres, classes=classes, agnostic=agnostic_nms)
+
+        # Process detections
+        for i, det in enumerate(pred):  # detections per image
+            if len(det):
+                # Rescale boxes from img_size to img0s size
+                det[:, :4] = scale_coords(img.shape[2:], det[:, :4], img0s.shape).round()
+
+                # Write results
+                for *xyxy, conf, cls in reversed(det):
+                    #crop an image based on coordinates
+                    object_coordinates = [int(xyxy[0]),int(xyxy[1]),int(xyxy[2]),int(xyxy[3])]
+                    cropobj_bgr = img0s[int(xyxy[1]):int(xyxy[3]),int(xyxy[0]):int(xyxy[2])]
+                    cropobj_rgb = cv2.cvtColor(cropobj_bgr, cv2.COLOR_BGR2RGB)
+                    clase = names[int(cls)]
+                    list_cropobj.append([Image.fromarray(cropobj_rgb),int(cls)])
+                    
+    return list_cropobj
+
 
 def get_attributes(input_img):
-    access_token = str(os.environ.get('key'))
+    #access_token = str(os.environ.get('key'))
+    access_token = key
     processor = DonutProcessor.from_pretrained("AA-supply/donut-finetuned-cedula", use_auth_token=access_token)
     model = VisionEncoderDecoderModel.from_pretrained("AA-supply/donut-finetuned-cedula", use_auth_token=access_token)
 
@@ -25,8 +102,12 @@ def get_attributes(input_img):
     model.eval()
     model.to(device)
     
-    image = check_image(input_img)
-    pixel_values = processor(image, return_tensors="pt").pixel_values
+    images = check_image(input_img)
+    images = crop(weights="best.pt", files= images)
+    image_cedula = [img[0] for img in images if img[1]==0][0] #0 en 'img[1]==0' es el label de cedula, si se reemplaza por 1
+                                                              #entrega licencias
+    
+    pixel_values = processor(image_cedula, return_tensors="pt").pixel_values
     pixel_values = pixel_values.to(device)
     print(pixel_values.size())
     # prepare decoder inputs

best.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:00bc306cd607b5ec22b8b499b81689081e2b478dd712dfa832955dc6c37684b4
+size 162520891

requirements.txt

@@ -3,4 +3,5 @@ transformers
 numpy
 datasets
 pdf2image
-poppler-utils
+poppler-utils
+opencv-python