v3.0.0

What's Changed

• Introduced persistence of analyses across the app
• Fine-tuned logo detection model
• Added and fine-tuned ResNet50 model for vector embedding
• Implemented FAISS ANN search
• Collected rest of images
• Labelled images for three brands: Under Armour, North Face and New Balance and added them to the tree
• Connected the API to all three model components to serve the full analysis
• Enhanced filtering on overlapping boundary boxes, distance and confidence thresholds

PR's Merged

• Added models : image, analysis, bounding box and brand by @dhrvrc in #105
• Feat/add resnet by @Hannaxgy in #109
• Removing overlapping bounding boxes by @Hannaxgy in #110
• Faiss for storing and loading! by @liuphui in #124
• Search with faiss by @Hannaxgy in #126
• Gather logo images for new balance #13 by @temi-oye in #125
• Tree building and pickles by @liuphui in #127
• Feat: Add image cropping service by @sktylr in #108
• update naming and added explanation by @Hannaxgy in #132
• Fixed issue where index starts from 1 instead of 0 by @liuphui in #134
• add north face labels by @temi-oye in #136
• updated the search function by @Hannaxgy in #135
• Generate pickles by @Hannaxgy in #137
• Label under armour images #123 by @temi-oye in #138
• Label new balance images 0 -129 #122 by @temi-oye in #152
• Added labels for New Balance images 260-389 by @kellytu274 in #153
• Feat/build clusters by @liuphui in #154
• Add north face PICKLES!!! by @Hannaxgy in #155
• Add northface by @liuphui in #156
• Add rest of new balance labels by @sktylr in #157
• New bal labels by @Hannaxgy in #158
• CENTROIDDDDSSS for new balance by @Hannaxgy in #159
• Centroids for north face by @liuphui in #160
• CENTROIDS :( by @Hannaxgy in #161
• Add and build tree by @sktylr in #162
• Api/create brands in database by @sktylr in #163
• Add custom auth by @sktylr in #164
• Add endpoints to list brands by @sktylr in #165
• Add command to run full detection and inference by @sktylr in #167
• Add puma images by @sktylr in #169
• Add license by @sktylr in #178
• Api/persist analyses database by @dhrvrc in #166
• Minor updates to serialization of response by @sktylr in #180
• Add service and E2E tests by @sktylr in #181
• Gather nike images by @kellytu274 in #182
• Add test model command by @sktylr in #183
• filtering distances by @Hannaxgy in #184
• returning bbox that dont overlap by @Hannaxgy in #188
• Adjust confidence score by @liuphui in #190
• Add logo recognition service, support brand filtering from FAISS results by @dhrvrc in #189
• Add brands API documentation by @sktylr in #191
• Modify images endpoint to allow users to filter brands by @dhrvrc in #192
• Introduce enabled flag on brands by @sktylr in #193
• Modifying API-Images Documentation by @dhrvrc in #194

New Contributors

• @Hannaxgy made their first contribution in #109
• @kellytu274 made their first contribution in #153

Full Changelog: v2.0.0...v3.0.0

v2.0.0

What's Changed

• Tests for API were set up
• Users and registration and login endpoints were added
• Confidence threshold of Yolo model tweaked following empirical analysis
• Issues with LFS fixed
• Further images collected and split

PR's Merged

• Chore: Setup tests by @sktylr in #94
• Gather logo images for north face #14 by @temi-oye in #91
• Feat: Add users by @sktylr in #92
• Feat: Add user authentication tests by @sktylr in #93
• Feat: Add S3 storage by @sktylr in #96
• Gather logo images for reebok #19 by @temi-oye in #99
• Changed conf_thres to be set as 0.09 based on research by @liuphui in #98
• Bugfix: Stop using LFS for images by @sktylr in #102
• Chore: Hide LFS migration from blame by @sktylr in #104
• Bugfix: Fix arbitrary CI test failures by @sktylr in #103
• Split logos for north face #64 by @temi-oye in #100
• Gather logo images for lululemon #20 by @temi-oye in #106

Full Changelog: v1.0.0...v2.0.0

v1.0.0

What's Changed

• Added integration with Yolov7 pre-trained model for logo detection in images
• Added API endpoint to facilitate processing n images through the model
• Upgraded docker building and dependencies for model integration
• Additional images collected for the dataset

PRs Merged

• Feat: Update docker building for model requirements by @sktylr in #69
• Added 'images' app by @dhrvrc in #70
• Chore: Bump Django and OpenCV versions by @sktylr in #75
• Feat: Add make temporary directory context manager by @sktylr in #73
• collect under armour images by @temi-oye in #72
• Bugfix: Update deployment script for larger docker images by @sktylr in #78
• Update README with CI/CD status badges and database migration instructions by @sktylr in #79
• Added model for detection and processing by @liuphui in #80
• Feat: Run models in docker without API by @sktylr in #82
• Linked Images endpoint to the Model by @dhrvrc in #90

New Contributors

• @liuphui made their first contribution in #80

Full Changelog: v0.1.0-alpha...v1.0.0

v0.1.0-alpha

v0.1.0-alpha

Pre-release following the first sprint from January 22nd - February 18th

What's changed

• Created initial repository structure including dockerising the API
• Added a CI/CD pipeline to test on pull requests and deploy on pushes to main
• An initial version of our web scraper was introduced with images from adidas added to our dataset.

Pull Requests Merged

• Add configuration scaffolding by @sktylr in #1
• Migrate api to model repo by @sktylr in #24
• Feat: Add CI pipeline by @sktylr in #46
• Add git attributes configuration for main file types by @sktylr in #48
• Design a web scraper for assisting with logo image search #18 by @temi-oye in #49
• Feat: Add CD pipeline by @sktylr in #52
• Gather logo images for adidas #11 by @temi-oye in #56

New Contributors

• @temi-oye made their first contribution in #49

Full Changelog: https://github.com/ias-tcd/competitor-matching-model/commits/v0.1.0-alpha