updated CHANGELOG.md

All notable changes to this project will be documented in this file.

## [1.7.0] - 18.08.2025

## [1.7.0] - 20.08.2025

### Added

- **Enhanced Metadata Extraction System**: Complete migration from exifread to exiftool for robust metadata processing

- **Video Export Support**: Camtrap DP observations table now supports video resources with first occurrence frame extraction

- **Enhanced AI Pipeline**: Robust, unified AI pipeline for both Citizen Science and FTPS data uploads, featuring improved administration with rerun capabilities and comprehensive status tracking

- **AI Pipeline Status Tracking**: Added `ai_pipeline_status` field to ClassificationProjectCollection for better pipeline monitoring

- **Enhanced Blurring System**: Comprehensive blur functionality improvements

  - `blur_humans`, `blur_vehicles`, and `blur_backup` fields to ClassificationProject model

  - Backup mechanism for original images before blurring operations

  - Configurable blur settings per classification project

  - Radius scaling based on image dimensions for consistent blur quality

  - Metadata preservation during blur operations

  - Cloud storage support for blur operations (REMOTE_STORAGE compatibility)

### Changed

- **FTPS Upload System**: Complete overhaul of FTPS-based data ingestion workflow

- **Video Processing**: Enhanced video timestamp calculation using `video_timestamp_for_target_frame` utility

- **AI Pipeline Robustness**: Improved error handling and logging in AI classification tasks

- **Serializer Tables**: Enhanced video bbox processing with first occurrence frame selection and proper timestamp alignment

- **Blur Processing Architecture**: Major overhaul of image blurring functionality

  - Migrated blur settings from AIProvider.anonimize_classes to individual ClassificationProject fields

  - Enhanced blur task with robust backup and restore capabilities

  - Improved blur quality with adaptive radius scaling based on image size

  - Better error handling and resource cleanup in blur operations

  - Support for both local and remote storage backends in blur processing

- Default overwrite_confidence changed to `True` for AI classification approvals

### Removed

- **Legacy Upload Client**: Discontinued support for trapper-client

- **Deprecated Functions**: Removed `get_image_modification_date_from_exif` function from common tools

- **Legacy Resource Filtering**: Simplified resource collection queries by removing period-based date filtering

- **Default AI Model Dependencies**: Removed `default_ai_model` field and related form configurations

- **Blur Configuration Consolidation**: Removed legacy anonymization system

  - `anonimize_classes` field from AIProvider model

  - ANONIMIZE_HUMAN, ANONIMIZE_VEHICLE, ANONIMIZE_HUMAN_AND_VEHICLE constants

  - Centralized blur settings management in ClassificationProject

### Dependencies

- **Removed**: exifread (>=3.0,<3.1) Python library

- **API Changes**: Resource serialization no longer includes `name` field (previously sourced from `prefixed_name`)

- **AI Model Configuration**: Migration 0085 removes `default_ai_model` field from ClassificationProject

- **Pipeline Status**: Migration 0086 adds `ai_pipeline_status` field for better monitoring

- **Blur Settings Migration**:

  - Migration 0087: adds blur fields (`blur_humans`, `blur_vehicles`, `blur_backup`) to ClassificationProject

  - Migration 0088: migrates blur settings from AIProvider.anonimize_classes to ClassificationProject blur fields

  - Migration 0089: removes `anonimize_classes` field from AIProvider

## [1.6.0] - 20.07.2025

                self.storage.delete(name)

                # Save the blurred image to the cloud storage

                name = self.storage.save(

                self.storage.save(

                    name, SimpleUploadedFile(filename, img_io.getvalue(), mime_type)

                )

            else:

            # of whether there's a file or not

            if bool(ff):

                try:

                    self.blur_image(ff, bboxes, resource.mime_type)

                    if (

                    status = self.blur_image(ff, bboxes, resource.mime_type)

                    if status and (

                        ObservationType.HUMAN in self.blur_observation_types

                        or ObservationType.VEHICLE in self.blur_observation_types

                    ):

                        resource.save()

                except Exception as e:

                    logger.error(f"Failed to blur {ff.name}: {str(e)}")

                    pass

    def process_user_classification(self, user_classification):

        # assume that bounding boxes from classification are already blurred