splitting large files

modelhub/cv/FaceMarkerLBF/FaceMarkerLBF.py

@@ -0,0 +1,41 @@
+from pathlib import Path
+
+import cv2
+import numpy as np
+from xlib.image import ImageProcessor
+from xlib.file import SplittedFile
+
+class FaceMarkerLBF:
+    def __init__(self):
+        path = Path(__file__).parent / 'lbfmodel.yaml'
+        SplittedFile.merge(path, delete_parts=False)
+        
+        marker = self.marker = cv2.face.createFacemarkLBF()
+        marker.loadModel(str(path))
+
+    def extract(self, img : np.ndarray):
+        """
+        arguments
+
+         img    np.ndarray  HW,HWC,NHWC
+
+        returns
+
+         [N,68,2]
+        """
+        ip = ImageProcessor(img)
+
+        N,H,W,_ = ip.get_dims()
+
+        feed_img = ip.to_grayscale().to_uint8().get_image('NHWC')
+
+        lmrks_list = []
+        for n in range( max(1,N) ):
+            _, lmrks = self.marker.fit(feed_img[n], np.array([ [0,0,W,H] ]) )
+            lmrks = lmrks[0][0]
+
+            lmrks_list.append(lmrks)
+
+        return np.float32(lmrks_list)
+
+

modelhub/cv/__init__.py

@@ -0,0 +1 @@
+from .FaceMarkerLBF.FaceMarkerLBF import FaceMarkerLBF

modelhub/onnx/S3FD/S3FD.py

@@ -7,6 +7,7 @@ from xlib.image import ImageProcessor
 from xlib.onnxruntime import (InferenceSession_with_device, ORTDeviceInfo,
                               get_available_devices_info)
 
+from xlib.file import SplittedFile
 
 class S3FD:
 
@@ -19,6 +20,8 @@ class S3FD:
             raise Exception(f'device_info {device_info} is not in available devices for S3FD')
 
         path = Path(__file__).parent / 'S3FD.onnx'
+        SplittedFile.merge(path, delete_parts=False)
+        
         self._sess = sess = InferenceSession_with_device(str(path), device_info)
         self._input_name = sess.get_inputs()[0].name
 

modelhub/torch/S3FD/S3FD.py

@@ -6,6 +6,7 @@ import torch
 import torch.nn as nn
 import torch.nn.functional as F
 from xlib import math as lib_math
+from xlib.file import SplittedFile
 from xlib.image import ImageProcessor
 from xlib.torch import TorchDeviceInfo, get_cpu_device
 
@@ -15,9 +16,12 @@ class S3FD:
         if device_info is None:
             device_info = get_cpu_device()
         self.device_info = device_info
-
+        
+        path = Path(__file__).parent / 'S3FD.pth'
+        SplittedFile.merge(path, delete_parts=False)
+        
         net = self.net = S3FDNet()
-        net.load_state_dict( torch.load(str(Path(__file__).parent / 's3fd.pth')) )
+        net.load_state_dict( torch.load(str(path) ))
         net.eval()
 
         if not device_info.is_cpu():