synapse-sdk 1.0.0b7__tar.gz → 1.0.0b8__tar.gz

{synapse_sdk-1.0.0b7/synapse_sdk.egg-info → synapse_sdk-1.0.0b8}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: synapse-sdk
-Version: 1.0.0b7
+Version: 1.0.0b8
 Summary: synapse sdk
 Author-email: datamaker <developer@datamaker.io>
 License: MIT

{synapse_sdk-1.0.0b7 → synapse_sdk-1.0.0b8}/docs/features/converters/index.md

@@ -12,19 +12,20 @@ The Synapse SDK provides comprehensive data format conversion utilities for comp
 
 The converter system supports bidirectional conversion between:
 
-- **DM Format** - Synapse Data Manager's native annotation format
-- **COCO Format** - Microsoft Common Objects in Context format
+- **DM Format** - Synapse Data Manager's native annotation format (supports v1 ⟷ v2 migration)
+- **COCO Format** - Microsoft Common Objects in Context format  
 - **Pascal VOC Format** - Visual Object Classes XML format
 - **YOLO Format** - You Only Look Once text-based format
 
-All converters support both categorized datasets (with train/valid/test splits) and non-categorized datasets.
+All converters support both categorized datasets (with train/valid/test splits) and non-categorized datasets. Additionally, all converters now support single file conversion mode for processing individual files.
 
 ## Supported Annotation Types
 
 | Annotation Type | DM | COCO | Pascal VOC | YOLO |
 |----------------|----|----|-----------|------|
 | Bounding Boxes | ✅ | ✅ | ✅ | ✅ |
-| Polygons/Segmentation | ✅ | ✅ | ❌ | ✅ |
+| Polygons | ✅ | ✅ | ❌ | ✅ |
+| Segmentation | ✅ | ✅ | ✅ | ❌ |
 | Keypoints | ✅ | ✅ | ❌ | ✅ |
 | Classifications | ✅ | ✅ | ❌ | ❌ |
 
@@ -35,7 +36,7 @@ All converters support both categorized datasets (with train/valid/test splits)
 Converts DM format annotations to Pascal VOC XML format.
 
 **Features:**
-- Converts bounding box annotations only
+- Converts bounding box annotations and segmentation masks
 - Creates standard Pascal VOC directory structure
 - Generates `classes.txt` file automatically
 - Supports both categorized and non-categorized datasets
@@ -59,6 +60,13 @@ converter = FromDMToPascalConverter(
 )
 converted_data = converter.convert()
 converter.save_to_folder('/output/pascal/dataset')
+
+# Single file conversion
+converter = FromDMToPascalConverter(is_single_conversion=True)
+with open('data.json') as f:
+    dm_data = json.load(f)
+with open('image.jpg', 'rb') as img_file:
+    pascal_xml = converter.convert_single_file(dm_data, img_file)
 ```
 
 **Input Structure (Categorized):**
@@ -101,7 +109,7 @@ Converts Pascal VOC XML annotations to DM format.
 **Features:**
 - Parses Pascal VOC XML files
 - Flexible directory naming (supports Annotations/annotations, Images/images/JPEGImages)
-- Extracts bounding box annotations only
+- Extracts bounding box annotations and segmentation information
 - Maintains class information
 
 **Usage:**
@@ -140,6 +148,13 @@ converter = FromDMToYOLOConverter(
 )
 converted_data = converter.convert()
 converter.save_to_folder('/output/yolo/dataset')
+
+# Single file conversion
+converter = FromDMToYOLOConverter(is_single_conversion=True)
+with open('data.json') as f:
+    dm_data = json.load(f)
+with open('image.jpg', 'rb') as img_file:
+    yolo_labels = converter.convert_single_file(dm_data, img_file)
 ```
 
 **Output Structure:**
@@ -200,7 +215,7 @@ Converts DM format to COCO format with full metadata support.
 
 **Features:**
 - Comprehensive COCO metadata (info, licenses, categories)
-- Supports bounding boxes, segmentation, and keypoints
+- Supports bounding boxes, polygons, segmentation, and keypoints
 - Dynamic category management
 - Extensible for different data types
 
@@ -235,6 +250,16 @@ converter = FromDMToCOCOConverter(
     licenses_list=licenses_list,
     is_categorized_dataset=True
 )
+
+# Single file conversion
+converter = FromDMToCOCOConverter(
+    data_type='img',
+    is_single_conversion=True
+)
+with open('data.json') as f:
+    dm_data = json.load(f)
+with open('image.jpg', 'rb') as img_file:
+    coco_annotation = converter.convert_single_file(dm_data, img_file)
 ```
 
 **Output Structure:**
@@ -347,17 +372,24 @@ with open('dm_v1_data.json', 'w') as f:
 All converters share these common parameters:
 
 ### `root_dir` (str)
-Path to the root directory containing the dataset.
+Path to the root directory containing the dataset. Not required when using single file conversion mode.
 
 ### `is_categorized_dataset` (bool)
 - `True`: Dataset has train/valid/test splits in separate subdirectories
 - `False`: Dataset is in a single directory without splits
 
+### `is_single_conversion` (bool)
+- `True`: Enable single file conversion mode for processing individual files
+- `False`: Process entire dataset directories (default behavior)
+
 ### Common Methods
 
 #### `convert()`
 Performs in-memory conversion and returns the converted data structure.
 
+#### `convert_single_file(data, original_file, **kwargs)`
+Available when `is_single_conversion=True`. Converts a single data object and corresponding original file.
+
 #### `save_to_folder(output_dir)`
 Saves the converted data to the specified output directory, creating the appropriate file structure for the target format.
 
@@ -399,4 +431,25 @@ coco_data = coco_converter.convert()
 coco_converter.save_to_folder('/data/coco_eval')
 ```
 
+### Single File Conversion Example
+
+```python
+import json
+from synapse_sdk.utils.converters.yolo.from_dm import FromDMToYOLOConverter
+
+# Initialize converter for single file processing
+converter = FromDMToYOLOConverter(is_single_conversion=True)
+
+# Load DM format data
+with open('annotation.json', 'r') as f:
+    dm_data = json.load(f)
+
+# Convert single file
+with open('image.jpg', 'rb') as img_file:
+    yolo_labels = converter.convert_single_file(dm_data, img_file)
+
+# yolo_labels contains the converted YOLO format labels
+print(yolo_labels)
+```
+
 This converter system provides a complete solution for dataset format transformation, enabling seamless integration between different machine learning workflows and annotation tools.

{synapse_sdk-1.0.0b7 → synapse_sdk-1.0.0b8}/synapse_sdk/plugins/categories/export/actions/export.py

@@ -200,7 +200,7 @@ class TaskExportTargetHandler(ExportTargetHandler):
         return value
 
     def get_results(self, client: Any, filters: dict):
-        filters['expand'] = 'data_unit'
+        filters['expand'] = ['data_unit', 'assignment', 'workshop']
         return client.list_tasks(params=filters, list_all=True)
 
     def get_export_item(self, results):
@@ -319,6 +319,7 @@ class ExportAction(Action):
                 first_result = next(peek_iter)  # Peek first value only
                 project_pk = first_result['project']
                 project_info = self.client.get_project(project_pk)
+                self.params['project_id'] = project_pk
                 self.params['configuration'] = project_info.get('configuration', {})
                 self.params['results'] = main_iter  # Keep original generator intact
             except (StopIteration, KeyError):

{synapse_sdk-1.0.0b7 → synapse_sdk-1.0.0b8/synapse_sdk.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: synapse-sdk
-Version: 1.0.0b7
+Version: 1.0.0b8
 Summary: synapse sdk
 Author-email: datamaker <developer@datamaker.io>
 License: MIT