fix serving bug of picodet

deploy/paddleserving/recognition/recognition_web_service.py

@@ -23,6 +23,7 @@ import faiss
 import pickle
 import json
 
+
 class DetOp(Op):
     def init_op(self):
         self.img_preprocess = Sequential([
@@ -65,22 +66,28 @@ class DetOp(Op):
 
             imgs.append({
                 "image": im[np.newaxis, :],
-              "im_shape": np.array(list(im.shape[1:])).reshape(-1)[np.newaxis,:],
-              "scale_factor": np.array([im_scale_y, im_scale_x]).astype('float32'),
+                "im_shape":
+                np.array(list(im.shape[1:])).reshape(-1)[np.newaxis, :],
+                "scale_factor":
+                np.array([im_scale_y, im_scale_x]).reshape(-1)[np.newaxis, :],
             })
         self.raw_img = raw_imgs
 
         feed_dict = {
-            "image":        np.concatenate([x["image"] for x in imgs], axis=0),
-            "im_shape":     np.concatenate([x["im_shape"] for x in imgs], axis=0),
-            "scale_factor": np.concatenate([x["scale_factor"] for x in imgs], axis=0)
+            "image": np.concatenate(
+                [x["image"] for x in imgs], axis=0),
+            "im_shape": np.concatenate(
+                [x["im_shape"] for x in imgs], axis=0),
+            "scale_factor": np.concatenate(
+                [x["scale_factor"] for x in imgs], axis=0)
         }
         return feed_dict, False, None, ""
 
     def postprocess(self, input_dicts, fetch_dict, log_id):
         boxes = self.img_postprocess(fetch_dict, visualize=False)
-        boxes.sort(key = lambda x: x["score"], reverse = True)
-        boxes = filter(lambda x: x["score"] >= self.threshold, boxes[:self.max_det_results])
+        boxes.sort(key=lambda x: x["score"], reverse=True)
+        boxes = filter(lambda x: x["score"] >= self.threshold,
+                       boxes[:self.max_det_results])
         boxes = list(boxes)
         for i in range(len(boxes)):
             boxes[i]["bbox"][2] += boxes[i]["bbox"][0] - 1
@@ -89,11 +96,12 @@ class DetOp(Op):
         res_dict = {"bbox_result": result, "image": self.raw_img}
         return res_dict, None, ""
 
+
 class RecOp(Op):
     def init_op(self):
         self.seq = Sequential([
-            BGR2RGB(), Resize((224, 224)), 
-            Div(255), Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225],
+            BGR2RGB(), Resize((224, 224)), Div(255),
+            Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225],
                       False), Transpose((2, 0, 1))
         ])
 
@@ -121,7 +129,8 @@ class RecOp(Op):
         origin_img = cv2.imdecode(data, cv2.IMREAD_COLOR)
         dt_boxes = input_dict["bbox_result"]
         boxes = json.loads(dt_boxes)
-        boxes.append({"category_id": 0,
+        boxes.append({
+            "category_id": 0,
             "score": 1.0,
             "bbox": [0, 0, origin_img.shape[1], origin_img.shape[0]]
         })
@@ -131,14 +140,14 @@ class RecOp(Op):
         imgs = []
         for box in boxes:
             box = [int(x) for x in box["bbox"]]
-            im = origin_img[box[1]: box[3], box[0]: box[2]].copy()
+            im = origin_img[box[1]:box[3], box[0]:box[2]].copy()
             img = self.seq(im)
             imgs.append(img[np.newaxis, :].copy())
 
         input_imgs = np.concatenate(imgs, axis=0)
         return {"x": input_imgs}, False, None, ""
 
-    def nms_to_rec_results(self, results, thresh = 0.1):
+    def nms_to_rec_results(self, results, thresh=0.1):
         filtered_results = []
         x1 = np.array([r["bbox"][0] for r in results]).astype("float32")
         y1 = np.array([r["bbox"][1] for r in results]).astype("float32")
@@ -187,12 +196,14 @@ class RecOp(Op):
         results = self.nms_to_rec_results(results, self.rec_nms_thresold)
         return {"result": str(results)}, None, ""
 
+
 class RecognitionService(WebService):
     def get_pipeline_response(self, read_op):
         det_op = DetOp(name="det", input_ops=[read_op])
         rec_op = RecOp(name="rec", input_ops=[det_op])
         return rec_op
 
+
 product_recog_service = RecognitionService(name="recognition")
 product_recog_service.prepare_pipeline_config("config.yml")
 product_recog_service.run_service()