Merge pull request #19 from rootvisionai/fix/labelImg_eksen

fix/labelImg_eksen

.gitignore

@@ -17,4 +17,5 @@
 *.onnx
 *.drawio
 /dev_gitignored/
-/frontend_python/
+!frontend_python/make_request_local.py
+

backend/server.py

@@ -215,7 +215,7 @@ def generate(gen_type):
                             if len(pts_) >= 3:
                                 print(f"---> Finding bounding box: {cnt_p}/{len(points_)}")
                                 pts = np.array([np.array(pt) for pt in pts_])
-                                y0, y1, x0, x1 = pts[:, 0].min(), pts[:, 0].max(), pts[:, 1].min(), pts[:, 1].max()
+                                x0, x1 ,y0, y1= pts[:, 0].min(), pts[:, 0].max(), pts[:, 1].min(), pts[:, 1].max()
                                 bboxes.append({
                                     "coordinates": [int(x0), int(y0), int(x1), int(y1)],
                                     "format": "xyxy",

config.yml

@@ -10,7 +10,7 @@ labeling:
   bounding_box: 1
   polygon_resolution: 0.3
 
-window_size: [1024, 1024]
+window_size: [512, 512]
 model: vit_h # Faster <<< vit_b, vit_l, vit_h >>> More accurate
 threshold: 0.9
 device: "cuda"

frontend_python/interface.py

@@ -80,15 +80,16 @@ def handle_click(event, x_click, y_click, flags, param):
 
 if __name__ == "__main__":
 
-
+    cfg = utils.load_config("./config.yml")
     annotations = []
     encoded_images = []
     image_paths = []
-    for image_id, image_path in enumerate(glob.glob(os.path.join("..", "support_images", "*.jpg"))):
+    for image_id, image_path in enumerate(glob.glob(os.path.join(".", "support_images", "*.jpg"))):
         init_image = utils.import_image(image_path)
         image_shape = init_image.shape
-        window_size = (512, 512)
-        image = cv2.resize(copy.deepcopy(init_image), (window_size[0], window_size[1]))
+        ref_width  = cfg.window_size[0] 
+        new_height = int((image_shape[0] / image_shape[1]) * ref_width)  
+        image = cv2.resize(copy.deepcopy(init_image), (ref_width,new_height))
         while True:
             print("CLICK ON POSITIVE POINTS")
             points = click_on_point(img=image)

frontend_python/interface_utils.py

@@ -2,6 +2,9 @@
 import numpy as np
 import base64
 import io
+import yaml
+import json
+import types
 
 def import_image(path):
     image_ = Image.open(path)
@@ -42,3 +45,13 @@ def get_image(image_data):
     image = Image.new("RGB", image_.size)
     image.paste(image_)
     return image
+
+def load_config(path_to_config_yaml="./config.yaml"):
+    with open(path_to_config_yaml) as f:
+        dct = yaml.safe_load(f)
+
+    def load_object(dct):
+        return types.SimpleNamespace(**dct)
+
+    cfg = json.loads(json.dumps(dct), object_hook=load_object)
+    return cfg

frontend_python/make_request_local.py

@@ -0,0 +1,82 @@
+import requests
+import json
+import interface_utils as utils
+import time, os
+import glob
+
+
+# Define the URL of the endpoint: http://fewshotsam.rootvisionai.net
+base = "localhost:8080"
+url = f"http://{base}/extract_features"  # replace with your actual endpoint
+
+# make get request
+if "http" in url:
+    response = os.system(f"curl --request GET http://{base}/health")
+    print(response)
+
+# Convert json to dict
+with open("request_content.json", "r") as fp:
+    data_json = json.load(fp)
+
+# Set the headers for the request
+headers = {
+    "Content-Type": "application/json"
+}
+
+# Send the POST request
+t0 = time.time()
+response = requests.post(url, data=json.dumps(data_json), headers=headers)
+
+t1 = time.time()
+print(response.text)
+data_json = json.loads(response.text)
+
+# Query Images: Load, extract, match
+cfg_data_result_dir = "./results/"
+cfg_data_query_dir = "./query_images/"
+cfg_data_format = "jpg"
+q_image_paths = glob.glob(os.path.join(cfg_data_query_dir, f"*.{cfg_data_format}"))
+q_image_paths += glob.glob(os.path.join(cfg_data_query_dir, "**", f"*.{cfg_data_format}"))
+
+for cnt, qip in enumerate(q_image_paths):
+    image_path = qip
+    image_filename = os.path.basename(image_path).split(".")[0]
+    data_json["image_path"] = image_path
+    init_image = utils.import_image(image_path)
+    data_json["image"] = utils.numpy_to_base64(init_image)
+
+    # Print the response
+    try:
+        print(data_json["error"] if "error" in data_json.keys() else data_json.keys())
+    except:
+        print(data_json)
+
+    # Define the URL of the endpoint
+    url = f"http://{base}/generate/all"
+
+    # Send the POST request
+    t1 = time.time()
+    response = requests.post(url, data=json.dumps(data_json), headers=headers)
+
+    t2 = time.time()
+    data_json = json.loads(response.text)
+    masks = utils.get_image(data_json["masks"])
+    masks.save(cfg_data_result_dir+f"masks_{image_filename}.png")
+
+    # Print the response
+    try:
+        print(data_json["error"] if "error" in data_json.keys() else data_json.keys())
+    except:
+        print(data_json)
+
+    # save annotation
+    with open(cfg_data_query_dir+f"{image_filename}.json", "w") as fp:
+        json.dump(data_json["coco_json"], fp, indent=4)
+
+
+    # save pascal_xml  
+    with open(cfg_data_query_dir+f"{image_filename}.xml", "w", encoding="utf-8") as fp:  
+        fp.write(str(data_json["pascal_xml"]))  
+
+    print(f"Request.1 Time: {t1-t0} | Request.2 Time: {t2-t1}")
+

version.txt

@@ -1 +1 @@
-0.0.3
+0.0.4