Merge Model-ROS into master #58
Open
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…ding Box Ratio Cost Functions * implement bbox ratio cost function * implement euclidean bbox ratio cost function * invert similarity to return cost instead
…unctions * Finished Class Gate cost matrix and weighted mean cost matrix functions * Improved Matrix Multiplication and Prevented Re-applying Cost Filter to Matrices * Corrected comment to match the research paper. We set the cost to 0 on mismatch
* Implemented Euclidean Cost and IOU_Euclidean cost functions * MTB-7 * MTB-8 updated euclidean cost for vectorization and returns ndarray
…aligned with shape (D, P) * Finished Class Gate cost matrix and weighted mean cost matrix functions * Improved Matrix Multiplication and Prevented Re-applying Cost Filter to Matrices * Corrected comment to match the research paper. We set the cost to 0 on mismatch * Updated the Weighted Mean Cost/Class Gate Cost Matrix functions to create a cost matrix of shape (num_detections, num_tracks)
* AVG COST MATRIX * Updated Function --------- Co-authored-by: Camila Barbosa <camilabarbosa@Camilas-MacBook-Air-3.local> Co-authored-by: David <78251284+david-orjuela@users.noreply.github.com>
* [MTB 1: IoU Cost Matrix] * Updated * UpdatedV2 * Updated V3 * UPDATED V4 --------- Co-authored-by: David <78251284+david-orjuela@users.noreply.github.com>
* [MTB 1: IoU Cost Matrix] * Updated * UpdatedV2 * Updated V3 * UPDATED V4 * Updated V5 * Updated V6 --------- Co-authored-by: David <78251284+david-orjuela@users.noreply.github.com>
* [MTB 1: IoU Cost Matrix] * Updated * UpdatedV2 * Updated V3 * UPDATED V4 * Updated V5 * Updated V6 --------- Co-authored-by: David <78251284+david-orjuela@users.noreply.github.com>
* [MTB 1: IoU Cost Matrix] * Updated * UpdatedV2 * Updated V3 * UPDATED V4 * Updated V5 * Updated V6 * Update V7.1 --------- Co-authored-by: David <78251284+david-orjuela@users.noreply.github.com>
…nters for more stable tracking
Mistaken creation
…use (detections, tracks) ordering
This commit introduces the foundational structure for training person Re-Identification models. The primary goal is to establish a verifiable and modular pipeline using a simple CNN and the Market-1501 dataset before introducing more complex backbones and metric learning. - Creates Modular Data Loader (`dataset.py`) - Creates Training Orcehstrator (`train.py`)
- Ignore large .mov and .mp4 files from demo
Co-authored-by: Jalyn N George <JalynG12563@users.noreply.github.com>
- `EmbeddedDetection2D.msg`: msg for `kestrel_vision`, containing each detection and its bounding box, class, confidence score, and appearance vector derived from CNN. - `EmbeddedDetection2DArray.msg`: The message sent out of `kestrel_vision`. It is an array containing all detections (array of above msg). - Removed `ObjectTrack.msg` -- outdated detection message. - Updated `CMakeLists` and `package.xml` to reflect these changes (removed `vision_msgs` dependencies) `kestrel_tracking` Updates - Moved `DataAssociation.py` here for use in Cost Matrix - Imported `tracking_node.py` from `model-team` branch (contains latest Camila/Gerard/Jalyn progress) - Removed `vision_msgs` dependencies from `package.xml` `kestrel_vision` Updates - `vision_msgs` Removal: Updated `package.xml` dependencies Removed import lines in `vision_node.py`, and assignment code. - Imported `best_model_checkpoint.pth` -- the best CNN weights from the last training -- for use in `vision_node.py`. Will improve. - Added corresponding model-weight-loading code
…acts (runs/tfevents/tb_env/checkpoints)
Co-authored-by: Gerard Gimenez <GerardGz@users.noreply.github.com>
…acking and kestrel_vision to match working local version.
- Removed duplicated track management code - Created a detection struct to hold detection data and avoid unnecessary recomputation (xyxy to center, to width, etc) - Updated Kalman filter state to use [cx, cy, w, h] measurements directly - Cleaned comments and removed typos - Added type hints throughout to catch data mismatched early - Added placeholder motion_cost for future data association integration Happy new year's eve!
…ation motion cost - Refactored TrackManager cost construction to fully matrix-based (T×D) motion/appearance costs - Standardized bbox convention to cxcywh for motion/gating and aligned DA outputs to tracker shape - Added lightweight adapters to stack Track/Det state and embeddings into NumPy arrays - Implemented vectorized Mahalanobis gating mask and cosine-distance appearance cost matrix
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.