YoloNAS-R #1983
Closed
YoloNAS-R #1983
Conversation
This file contains 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
…ygon-polygon iou for metric
# Conflicts: # src/super_gradients/training/utils/distributed_training_utils.py
…#1994) * Split processing and default preprocesisng params into separate files * Put back method for BC support * Fix imports * Revert sampelr
…feature/SG-1448-OBB # Conflicts: # src/super_gradients/training/processing/__init__.py # src/super_gradients/training/processing/defaults.py # src/super_gradients/training/processing/processing.py
* OBB loss * DOTA dataset * Adding loss, postprocessing & visualization * Visualization callback * Visualization callback * YoloNAS-R * YoloNAS-R * YoloNAS-R * optimized_rboxes_nms * optimized_rboxes_nms * optimized_rboxes_nms * optimized_rboxes_nms * optimized_rboxes_nms * Rename variables for better clarity * Optimize loss weights * Prepare data script * Prepare data script * Reduce batch size & topk * Remove unused arg * Remove unused arg * Remove unused arg * Remove unused arg * Increase topk * Increase topk * Increase topk * Increase topk * Increase topk * Increase topk * Increase topk * Increase topk * Increase topk * Increase topk * Increase topk * Increase topk * Increase topk * Increase topk * RAdam * RAdam * RAdam * CIou=False * Disable check_points_inside_rboxes * Disable check_points_inside_rboxes * Disable check_points_inside_rboxes * Tune weights * set_anomaly_enabled * set_anomaly_enabled * Increase eps * Multiply by scalar * Multiply by scalar * average_losses_in_ddp * yolo_nas_r_balanced * Limit angle * Optimize image slicing * Do not reduce losses, reduce only scores sum * Do not reduce losses, reduce only scores sum * Do not reduce losses, reduce only scores sum * Clip grad * Disable L1 components * Disable L1 components * Disable L1 components * Disable L1 components * predict() support for YoloNAS-R * Comment saving of visualization * yolo_nas_r_tzag * Remove parameter * Fixed newline * Increase min confidence * Added missing ReverseImageChannels * yolo_nas_r_tzag_balanced * Added OBB transforms * Added OBB transforms * get_dataset_preprocessing_params * Added removal of small boxes during training * Added removal of small boxes during training * Added removal of invalid boxes * Added augmentations * Increase bs * Ensure that poly_to_cxcywhr always return boxes with w > h * Enabled anomaly detection * Added eps * Undo anomaly * dota_yolo_nas_r_balanced_no_mixup * Added logging of non-finite IoU results * Added sanitize sample call after applying albumentations transforms * Replaced atan2() with atan() in CIoU loss because gt boxes can be zeros * MOAR Augs * MOAR Augs * RAdam -> AdamW Less augs * dota_yolo_nas_r_balanced_pretrain * Tune augs * Fixed issue of logging wrong config * Revert file * Cleanup * Added docs * Remove non-existing params * YoloNAS-R M&L variants * Export support for YoloNAS R * Temporary comment matrix nms * PrefetchIterator * Remove gs import * dota_yolo_nas_r_s * Increase BS * Enable anomaly detection * Rewrite t3 * Rewrite t3 * Rewrite t3 * dota_yolo_nas_r_s_1_gpu * M & L * Increase numerical stability * Remove prefetch * Increase BS for L * Increase BS for L * Disable fp16 * RM anomaly * RM anomaly * fp32 * Increase batch size * Increase batch size * Increase batch size * dota_yolo_nas_r_l_fp32 * Switch to use of matrix nms in post-prediction callback and exact polygon-polygon iou for metric * Switch to use of matrix nms in post-prediction callback and exact polygon-polygon iou for metric * Added docs for rboxes_matrix_nms * Update script * Improve auto-generated submission name * Added positional args * Added positional args * dota_yolo_nas_r_l_fp32_mixup * dota_yolo_nas_r_l_fp32_mixup * dota_yolo_nas_r_l_fp32_mixup * dota_yolo_nas_r_l_fp32_mixup * Update recipes * Update recipes * Added license support for YoloNAS-R * Improve docstrings * Remove max gradient debugging * Improve docstrings * Remove random rotate90 * Move OBBSample to samples * Added test
…feature/SG-1448-OBB # Conflicts: # src/super_gradients/training/datasets/data_formats/obb/cxcywhr.py
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.
This is a draft PR
What in the box:
What is missing