Allow relative pathnames for importing/exporting annotations

README.imports

@@ -0,0 +1,7 @@
+Importing from other sources of COCO annotations can be very cumbersome, often relative pathnames are encouraged (in the coco-annotator too apparently), however importing or exporting removes that relationship.
+A new attribute has been added to the ImageCatalouge to capture the relative path of an image at import time as well as scanning. if there are no sub-directories this variable will be equal to the file_name, however the search logic will use the relative pathname to discover the image.
+When exporting the relative path will replace the 'file_name' entry in the 'image' dictionary, this behaviour can be switched off by setting the 'EXPORT_RELPATH' environment variable to false (in the compose file for the workers).
+If the annotation only contains a bbox and no valid segmentation the segmentation is derived form the bbox. Previously this annotation would have been ignored.
+
+ Uwe Rosebrock July 2021
+

backend/config/config.py

@@ -50,6 +50,8 @@ class Config:
     WORKER_CONNECTIONS = 1000
 
     TESTING = os.getenv("TESTING", False)
+
+    EXPORT_RELPATH = os.getenv("EXPORT_RELPATH", True)
 
     ### Workers
     CELERY_BROKER_URL = os.getenv("CELERY_BROKER_URL", "amqp://user:password@messageq:5672//")

backend/database/images.py

@@ -1,22 +1,22 @@
 import os
 import imantics as im
 
-
+from pathlib import PurePath
 from PIL import Image, ImageFile
 from mongoengine import *
 
 from .events import Event, SessionEvent
 from .datasets import DatasetModel
 from .annotations import AnnotationModel
-
+from config import Config
 
 ImageFile.LOAD_TRUNCATED_IMAGES = True
 
 
 class ImageModel(DynamicDocument):
 
     COCO_PROPERTIES = ["id", "width", "height", "file_name", "path", "license",\
-                       "flickr_url", "coco_url", "date_captured", "dataset_id"]
+                       "flickr_url", "coco_url", "date_captured", "dataset_id","relpath","num_annotations"]
 
     # -- Contants
     THUMBNAIL_DIRECTORY = '.thumbnail'
@@ -35,6 +35,7 @@ class ImageModel(DynamicDocument):
 
     # Absolute path to image file
     path = StringField(required=True, unique=True)
+    relpath = StringField(required=True, unique=True)
     width = IntField(required=True)
     height = IntField(required=True)
     file_name = StringField()
@@ -74,16 +75,21 @@ def create_from_path(cls, path, dataset_id=None):
 
         if dataset_id is not None:
             image.dataset_id = dataset_id
+            dataset = DatasetModel.objects.get(id=dataset_id)
+            directory = os.path.join(Config.DATASET_DIRECTORY, dataset.name)
         else:
             # Get dataset name from path
             folders = path.split('/')
+
             i = folders.index("datasets")
             dataset_name = folders[i+1]
-
+            directory = os.path.join(*folders[:i+2])
             dataset = DatasetModel.objects(name=dataset_name).first()
             if dataset is not None:
                 image.dataset_id = dataset.id
-
+
+        # UR added relpath
+        image.relpath = str(PurePath(image.path).relative_to(directory))
         pil_image.close()
 
         return image

backend/database/tasks.py

@@ -40,6 +40,10 @@ class TaskModel(DynamicDocument):
     _update_every = 10
     _progress_update = 0
 
+
+    def debug(self, string):
+        self._log(string, level="DEBUG")
+
     def error(self, string):
         self._log(string, level="ERROR")
 

backend/workers/tasks/data.py

@@ -18,8 +18,14 @@
 from celery import shared_task
 from ..socket import create_socket
 from mongoengine import Q
+from config import Config
+from pathlib import PurePath
 
 
+def bbox2seg(bbox):
+    return [bbox[0],bbox[1],bbox[0]+bbox[2],bbox[1],bbox[0]+bbox[2],bbox[1]+bbox[3],bbox[0],bbox[1]+bbox[3]]
+
+
 @shared_task
 def export_annotations(task_id, dataset_id, categories):
 
@@ -74,7 +80,10 @@ def export_annotations(task_id, dataset_id, categories):
     total_images = db_images.count()
     for image in db_images:
         image = fix_ids(image)
-
+
+        if Config.EXPORT_RELPATH and 'relpath' in image:
+            image['file_name'] = image['relpath']
+
         progress += 1
         task.set_progress((progress / total_items) * 100, socket=socket)
 
@@ -103,7 +112,11 @@ def export_annotations(task_id, dataset_id, categories):
 
                 num_annotations += 1
                 coco.get('annotations').append(annotation)
-
+        '''
+        if num_annotations > 0:
+            image["num_annotations"]=num_annotations
+            image["annotated"]=True
+        '''
         task.info(
             f"Exporting {num_annotations} annotations for image {image.get('id')}")
         coco.get('images').append(image)
@@ -135,7 +148,8 @@ def import_annotations(task_id, dataset_id, coco_json):
 
     task = TaskModel.objects.get(id=task_id)
     dataset = DatasetModel.objects.get(id=dataset_id)
-
+    # UR added relpath
+    directory = os.path.join(Config.DATASET_DIRECTORY, dataset.name)
     task.update(status="PROGRESS")
     socket = create_socket()
 
@@ -203,12 +217,12 @@ def import_annotations(task_id, dataset_id, coco_json):
     for image in coco_images:
         image_id = image.get('id')
         image_filename = image.get('file_name')
-
+        
         # update progress
         progress += 1
         task.set_progress((progress / total_items) * 100, socket=socket)
-
-        image_model = images.filter(file_name__exact=image_filename).all()
+        # UR added relpath
+        image_model = images.filter(relpath=image_filename).all()
 
         if len(image_model) == 0:
             task.warning(f"Could not find image {image_filename}")
@@ -239,11 +253,11 @@ def import_annotations(task_id, dataset_id, coco_json):
         progress += 1
         task.set_progress((progress / total_items) * 100, socket=socket)
 
-        has_segmentation = len(segmentation) > 0
+        has_segmentation = (len(segmentation) > 0 or isbbox) and sum(bbox) > 1
         has_keypoints = len(keypoints) > 0
         if not has_segmentation and not has_keypoints:
             task.warning(
-                f"Annotation {annotation.get('id')} has no segmentation or keypoints")
+                f"Annotation {annotation.get('id')} has no segmentation, bbox or keypoints")
             continue
 
         try:
@@ -259,7 +273,8 @@ def import_annotations(task_id, dataset_id, coco_json):
             image_id=image_model.id,
             category_id=category_model_id,
             segmentation=segmentation,
-            keypoints=keypoints
+            keypoints=keypoints,
+            bbox = bbox
         ).first()
 
         if annotation_model is None:
@@ -272,9 +287,14 @@ def import_annotations(task_id, dataset_id, coco_json):
             annotation_model.metadata = annotation.get('metadata', {})
 
             if has_segmentation:
+                if len(segmentation) < 1 or len(segmentation[0]) < 1: ## we have an empty segment with a bbox
+                    task.info(f"Creating segment from bbox {bbox}")
+                    segmentation = [bbox2seg(bbox)]
+                    isbbox = True
+
                 annotation_model.segmentation = segmentation
-                annotation_model.area = area
-                annotation_model.bbox = bbox
+            annotation_model.area = area
+            annotation_model.bbox = bbox
 
             if has_keypoints:
                 annotation_model.keypoints = keypoints
@@ -283,6 +303,7 @@ def import_annotations(task_id, dataset_id, coco_json):
             annotation_model.save()
 
             image_categories.append(category_id)
+
         else:
             annotation_model.update(deleted=False, isbbox=isbbox)
             task.info(

backend/workers/tasks/scan.py

@@ -7,7 +7,7 @@
 from celery import shared_task
 from ..socket import create_socket
 from .thumbnails import thumbnail_generate_single_image
-
+from pathlib import PurePath
 import os
 
 
@@ -22,36 +22,42 @@ def scan_dataset(task_id, dataset_id):
 
     directory = dataset.directory
     toplevel = list(os.listdir(directory))
-    task.info(f"Scanning {directory}")
+    task.info(f"Scanning {directory} ")
 
     count = 0
     for root, dirs, files in os.walk(directory):
-
+        task.info(f"Scanning {directory} at {root}")
         try:
-            youarehere = toplevel.index(root.split('/')[-1])
-            progress = int(((youarehere)/len(toplevel))*100)
+            if root in toplevel:
+                youarehere = toplevel.index(root.split('/')[-1])
+                progress = int(((youarehere)/len(toplevel))*100)
+            else:
+                progress = len(toplevel)/100
+                youarehere = root
             task.set_progress(progress, socket=socket)
-        except:
-            pass
+        except Exception as ee:
+            task.warning(f"Could not set progress {youarehere} because of {ee}")
 
         if root.split('/')[-1].startswith('.'):
+            task.debug(f"Ignoring hidden root: {root}")
             continue
 
         for file in files:
             path = os.path.join(root, file)
-
+            relpath = str(PurePath(path).relative_to(directory))
             if path.endswith(ImageModel.PATTERN):
-                db_image = ImageModel.objects(path=path).first()
+                db_image = ImageModel.objects(relpath=relpath).first()
 
                 if db_image is not None:
+                    task.debug(f"File already exists: {relpath}")
                     continue
 
                 try:
                     ImageModel.create_from_path(path, dataset.id).save()
                     count += 1
                     task.info(f"New file found: {path}")
-                except:
-                    task.warning(f"Could not read {path}")
+                except Exception as e:
+                    task.warning(f"Could not read {path} because of {e}")
 
     [thumbnail_generate_single_image.delay(image.id) for image in ImageModel.objects(regenerate_thumbnail=True).all()]
 

build_local.sh

@@ -0,0 +1,16 @@
+#!/bin/bash
+if [[ "$1" == regi* ]];then
+
+docker build -t docker-registry.it.csiro.au/trike/uwer/coco-annotator:python-env -f ./backend/Dockerfile .
+docker build -t docker-registry.it.csiro.au/trike/uwer/annotator_webclient -f Dockerfile .
+docker build -t docker-registry.it.csiro.au/trike/uwer/annotator_workers -f ./backend/workers/Dockerfile .
+
+docker push docker-registry.it.csiro.au/trike/uwer/coco-annotator:python-env
+docker push docker-registry.it.csiro.au/trike/uwer/annotator_webclient
+docker push docker-registry.it.csiro.au/trike/uwer/annotator_workers
+else
+docker build -f ./backend/Dockerfile . -t uwer/coco-annotator:python-env 
+docker build . -t uwer/annotator_webserver
+docker build -f ./backend/workers/Dockerfile . -t uwer/annotator_workers 
+
+fi 

docker-compose-local.yml

@@ -0,0 +1,61 @@
+version: "3"
+volumes:
+  mongodb_data_uwer:
+    external: false
+
+networks:
+  cocoannotator_uwer:
+    external: false
+
+services:
+  webserver:
+    #image: uwer/coco-annotator
+    image: uwer/annotator_webserver
+    #:webserver-stable
+    container_name: annotator_webserver_uwer
+    restart: always
+    ports:
+      - "5050:5000"
+    environment:
+      - SECRET_KEY=RandomSecretKeyHere
+      - FILE_WATCHER=true
+    volumes:
+      - "./datasets:/datasets"
+      - "./models:/models"
+    depends_on:
+      - database
+      - workers
+    networks:
+      - cocoannotator_uwer
+  workers:
+    container_name: annotator_workers_uwer
+    image: uwer/annotator_workers
+    #:workers-stable
+    volumes:
+      - "./datasets:/datasets"
+    depends_on:
+      - messageq
+      - database
+    networks:
+      - cocoannotator_uwer
+  messageq:
+    image: rabbitmq:3
+    container_name: annotator_message_q_uwer
+    environment:
+      - RABBITMQ_DEFAULT_USER=user
+      - RABBITMQ_DEFAULT_PASS=password
+      - EXPORT_RELPATH=true
+    networks:
+      - cocoannotator_uwer
+  database:
+    image: mongo:4.0
+    container_name: annotator_mongodb_uwer
+    restart: always
+    environment:
+      - MONGO_DATA_DIR=/data/db
+      - MONGO_LOG_DIR=/dev/null
+    volumes:
+      - "mongodb_data_uwer:/data/db"
+    command: "mongod --smallfiles --logpath=/dev/null"
+    networks:
+      - cocoannotator_uwer