oriented-det 0.3.0__tar.gz → 0.3.1__tar.gz

{oriented_det-0.3.0 → oriented_det-0.3.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: oriented-det
-Version: 0.3.0
+Version: 0.3.1
 Summary: A lightweight, modern PyTorch library for rotated object detection in aerial and satellite imagery
 Author: Jeff Faudi, DL4EO
 Maintainer: Jeff Faudi, DL4EO
@@ -77,7 +77,7 @@ Dynamic: license-file
 - **Datasets**: DOTA polygon loader (pattern, split file, or separate folders), **HRSC2016** native XML loader, **FAIR1M** 37-class XML loader (+ convert/tile tutorial), **SSDD** and **HRSID** SAR ship loaders (finetune DOTA 1× Hub locally; no SAR zoo), image tiling, label filtering, ignore masks, oriented mAP evaluation
 - **Models**: **Oriented R-CNN** ([Xie et al., ICCV 2021](https://openaccess.thecvf.com/content/ICCV2021/html/Xie_Oriented_R-CNN_for_Object_Detection_ICCV_2021_paper.html); horizontal RPN + MidpointOffset → oriented RoIAlign + oriented ROI head), **Rotated Faster R-CNN** (Ren et al., NeurIPS 2015 two-stage baseline with horizontal RPN + horizontal RoIAlign + rotated ROI head; MMRotate reference), **Rotated RetinaNet** ([Lin et al., ICCV 2017](https://openaccess.thecvf.com/content_ICCV_2017/papers/Lin_Focal_Loss_for_ICCV_2017_paper.pdf); oriented anchors, sigmoid focal loss), **Rotated FCOS** (anchor-free single-stage; distance-angle coder, centerness, L1 / KFIoU / decoded rIoU); ResNet + FPN backbones; selective loading of external checkpoints where configs wire `checkpoint.load_from_checkpoint`
 - **Training**: JSON configs + **`odet train`**, mixed precision (AMP), gradient accumulation, checkpointing, best-metric tracking, TensorBoard, optional curriculum learning and profiling
-- **ONNX export**: pre-NMS ONNX + Python NMS (`python -m export` / `make export-onnx`); consumers run ONNX Runtime only
+- **ONNX export**: pre-NMS ONNX + Python NMS (`odet export` / `make export-onnx`); consumers run ONNX Runtime only
 
 Hands-on write-ups (code, metrics, inference knobs) live on **[DeepLearning.Earth](https://deeplearning.earth)** — especially the [15 DOTA class tour](https://deeplearning.earth/posts/2026-06-23_oriented_rcnn_detections_for_the_15_dota_classes/), [sampled rIoU and ProbIoU](https://deeplearning.earth/posts/2026-07-10_rotated_faster_rcnn_probiou_dota/), [Rotated FCOS / v0.2.0](https://deeplearning.earth/posts/2026-08-28_oriented-det_v0_2_0_rotated_fcos_decoded_riou_and_the_updated_zoo/), and [macOS FCOS vs Oriented R-CNN](https://deeplearning.earth/posts/2026-09-02_rotated_fcos_vs_oriented_rcnn_on_macos/).
 
@@ -107,7 +107,7 @@ uv pip install -e .
 - From PyPI: `pip install oriented-det`
 - For development and tests: `uv pip install -e ".[dev]"`
 - For the Gradio prediction viewer: `uv pip install -e ".[viewer]"` or `pip install "oriented-det[viewer]"`
-- For **ONNX export**: `uv pip install -e ".[export]"` then `python -m export --help` (see [ONNX export](docs/examples/export.md))
+- For **ONNX export**: `uv pip install -e ".[export]"` then `odet export --help` (see [ONNX export](docs/examples/export.md))
 - For **macOS Apple Silicon** or **CPU-only**, see [Installation](docs/getting-started/installation.md) and the [macOS walkthrough](https://deeplearning.earth/posts/2026-06-25_oriented_object_detection_on_macos_in_pure_python/) (`odet image-demo` on MPS, no CUDA toolchain).
 - Verify: `pytest tests/test_geometry.py tests/test_iou.py tests/test_nms.py`
 
@@ -154,7 +154,7 @@ After that, copy-pasted commands and unmodified configs that reference `/path/to
 | **Configs** | [`configs/`](configs/) | Experiment JSON (`_base_` inheritance, schema in `configs/config.schema.json`) |
 | **Runs** | `runs/<model_type>/<timestamp>/` | Checkpoints, `config.json` snapshot, `train.log` (created at train time; not shipped in the repo) |
 | **Docs** | [`docs/`](docs/) | MkDocs user guide and API reference |
-| **Examples** | [`demo/`](demo/), [`pretrained/`](pretrained/), [`notebooks/`](notebooks/), [`export/`](export/) | Demo images; Hub checkpoints; Kaggle FAIR1M tutorial notebook; ONNX export producer (`python -m export`) |
+| **Examples** | [`demo/`](demo/), [`pretrained/`](pretrained/), [`notebooks/`](notebooks/), [`export/`](export/) | Demo images; Hub checkpoints; Kaggle FAIR1M tutorial notebook; ONNX export producer (`odet export`) |
 
 **`odet` vs `tools/`:** Installing the package registers the `odet` command. It loads modules under `tools/` (for example `tools.train`, `tools.save_predictions`). Shared inference and collate code lives in [`oriented_det/runtime/`](oriented_det/runtime/). You do not need two workflows — use **`odet`** (or **`make`**, which calls `odet`).
 
@@ -165,7 +165,7 @@ After that, copy-pasted commands and unmodified configs that reference `/path/to
 Full documentation is in the **docs/** folder and can be built and served with MkDocs:
 
 - **Build/serve**: `make docs` or `make docs-serve` (see [docs/README.md](docs/README.md)); or `uv pip install -e ".[docs]"` then `mkdocs serve`.
-- **Guides**: [Getting Started](docs/getting-started/installation.md), [User Guide](docs/user-guide/geometry.md), [API Reference](docs/api/geometry.md), [Examples](docs/examples/inference.md), [ONNX export](docs/examples/export.md).
+- **Guides**: [Getting Started](docs/getting-started/installation.md), [User Guide](docs/user-guide/geometry.md), [API Reference](docs/api/geometry.md), [Examples](docs/examples/inference.md), [Docker deploy](docs/examples/deploy.md), [ONNX export](docs/examples/export.md).
 - **Blog**: [DeepLearning.Earth](https://deeplearning.earth) — curated posts below (code, metrics, inference knobs).
 
 | Topic | Post |
@@ -181,6 +181,8 @@ Full documentation is in the **docs/** folder and can be built and served with M
 | FCOS vs Oriented R-CNN (MPS latency, thresholds) | [https://deeplearning.earth/posts/2026-09-02_rotated_fcos_vs_oriented_rcnn_on_macos/](https://deeplearning.earth/posts/2026-09-02_rotated_fcos_vs_oriented_rcnn_on_macos/) |
 | Side-by-side browser demo (three 1× checkpoints) | [https://deeplearning.earth/posts/2026-09-06_oriented_det_optical_satellite_demo/](https://deeplearning.earth/posts/2026-09-06_oriented_det_optical_satellite_demo/) |
 | Apache 2.0 vs DOTA / HRSC dataset terms | [https://deeplearning.earth/posts/2026-09-10_oriented_det_apache_license_versus_dota/](https://deeplearning.earth/posts/2026-09-10_oriented_det_apache_license_versus_dota/) |
+| Docker deploy (Tile Geo Process) | [https://deeplearning.earth/posts/2026-10-05_deploy_oriented_det_in_docker/](https://deeplearning.earth/posts/2026-10-05_deploy_oriented_det_in_docker/) |
+| ONNX export (no PyTorch on infer) | [https://deeplearning.earth/posts/2026-10-08_onnx_export_without_pytorch/](https://deeplearning.earth/posts/2026-10-08_onnx_export_without_pytorch/) |
 
 ## Documentation by folder
 
@@ -191,8 +193,8 @@ Full documentation is in the **docs/** folder and can be built and served with M
 | [oriented_det/cli/](oriented_det/cli/) | [oriented_det/cli/README.md](oriented_det/cli/README.md) | **`odet`** entry point and subcommand list |
 | [tools/](tools/) | [tools/README.md](tools/README.md) | CLI script implementations (invoked by `odet`; see [Repository layout](#repository-layout)) |
 | [configs/](configs/) | [configs/README.md](configs/README.md) | DOTA configs and **pretrain model zoo** (`_base_` inheritance) |
-| [deploy/example/](deploy/example/) | [deploy/example/README.md](deploy/example/README.md) | Minimal DOTA deploy smoke image |
-| [export/](export/) | [export/README.md](export/README.md) | ONNX export producer (`python -m export`); consumer bundle in gitignored `onnx_export/` |
+| [deploy/example/](deploy/example/) | [deploy/example/README.md](deploy/example/README.md) | Sanic Tile Geo Process Docker image (DOTA smoke) |
+| [export/](export/) | [export/README.md](export/README.md) | ONNX export producer (`odet export`); consumer bundle in gitignored `onnx_export/` |
 | [docs/](docs/) | [docs/README.md](docs/README.md) | MkDocs source; full user guide and API reference |
 
 ## Training and evaluation
@@ -209,7 +211,7 @@ Install once: `uv pip install -e .`. Then:
 | COCO polygons → DOTA | `odet coco-to-dota --ann-file instances.json --image-dir images --output-dir out` |
 | Val predictions | `odet preds --experiment-dir runs/oriented_rcnn/<timestamp>` or `make preds` |
 | Offline mAP | `make eval-val` or `make preds` then `make metrics`. DOTA: leaky val tiles (real test is Task 1). HRSC/FAIR1M/SSDD/HRSID: held-out test/val. |
-| ONNX export | `make export-onnx` or `python -m export onnx ...` (see [export/README.md](export/README.md)) |
+| ONNX export | `odet export onnx ...` or `make export-onnx` (see [export/README.md](export/README.md)) |
 
 DOTA configs: per-model `dota_le90_1x.json` under [configs/](configs/) (3× where published; Faster R-CNN and FCOS DOTA are 1×). Run `odet --help` for all subcommands. Makefile shortcuts and script-level options: [tools/README.md](tools/README.md). Config reference: [docs/user-guide/configuration.md](docs/user-guide/configuration.md), [configs/config.schema.json](configs/config.schema.json), [configs/README.md](configs/README.md).
 
@@ -245,7 +247,7 @@ OrientedDet is Apache-2.0 open source. For **consulting, training workshops, or
 
 ## Publishing to PyPI
 
-Version **0.3.0** — tag releases as **`v0.3.0`** (git) matching `version` in `pyproject.toml`.
+Version **0.3.1** — tag releases as **`v0.3.1`** (git) matching `version` in `pyproject.toml`.
 
 Configs: edit **`configs/`** at the repo root, then **`make sync-configs`** so **`oriented_det/configs/`** stays in sync (see **`oriented_det/configs/vendored_manifest.txt`**). CI runs **`make check-configs`**.
 
@@ -276,7 +278,7 @@ make check-configs
 pytest tests/ -q
 ```
 
-Bump version in `pyproject.toml` and `docs/changelog.md` if needed (currently **0.3.0**).
+Bump version in `pyproject.toml` and `docs/changelog.md` if needed (currently **0.3.1**).
 
 #### 3. Build locally
 

{oriented_det-0.3.0 → oriented_det-0.3.1}/README.md

@@ -17,7 +17,7 @@
 - **Datasets**: DOTA polygon loader (pattern, split file, or separate folders), **HRSC2016** native XML loader, **FAIR1M** 37-class XML loader (+ convert/tile tutorial), **SSDD** and **HRSID** SAR ship loaders (finetune DOTA 1× Hub locally; no SAR zoo), image tiling, label filtering, ignore masks, oriented mAP evaluation
 - **Models**: **Oriented R-CNN** ([Xie et al., ICCV 2021](https://openaccess.thecvf.com/content/ICCV2021/html/Xie_Oriented_R-CNN_for_Object_Detection_ICCV_2021_paper.html); horizontal RPN + MidpointOffset → oriented RoIAlign + oriented ROI head), **Rotated Faster R-CNN** (Ren et al., NeurIPS 2015 two-stage baseline with horizontal RPN + horizontal RoIAlign + rotated ROI head; MMRotate reference), **Rotated RetinaNet** ([Lin et al., ICCV 2017](https://openaccess.thecvf.com/content_ICCV_2017/papers/Lin_Focal_Loss_for_ICCV_2017_paper.pdf); oriented anchors, sigmoid focal loss), **Rotated FCOS** (anchor-free single-stage; distance-angle coder, centerness, L1 / KFIoU / decoded rIoU); ResNet + FPN backbones; selective loading of external checkpoints where configs wire `checkpoint.load_from_checkpoint`
 - **Training**: JSON configs + **`odet train`**, mixed precision (AMP), gradient accumulation, checkpointing, best-metric tracking, TensorBoard, optional curriculum learning and profiling
-- **ONNX export**: pre-NMS ONNX + Python NMS (`python -m export` / `make export-onnx`); consumers run ONNX Runtime only
+- **ONNX export**: pre-NMS ONNX + Python NMS (`odet export` / `make export-onnx`); consumers run ONNX Runtime only
 
 Hands-on write-ups (code, metrics, inference knobs) live on **[DeepLearning.Earth](https://deeplearning.earth)** — especially the [15 DOTA class tour](https://deeplearning.earth/posts/2026-06-23_oriented_rcnn_detections_for_the_15_dota_classes/), [sampled rIoU and ProbIoU](https://deeplearning.earth/posts/2026-07-10_rotated_faster_rcnn_probiou_dota/), [Rotated FCOS / v0.2.0](https://deeplearning.earth/posts/2026-08-28_oriented-det_v0_2_0_rotated_fcos_decoded_riou_and_the_updated_zoo/), and [macOS FCOS vs Oriented R-CNN](https://deeplearning.earth/posts/2026-09-02_rotated_fcos_vs_oriented_rcnn_on_macos/).
 
@@ -47,7 +47,7 @@ uv pip install -e .
 - From PyPI: `pip install oriented-det`
 - For development and tests: `uv pip install -e ".[dev]"`
 - For the Gradio prediction viewer: `uv pip install -e ".[viewer]"` or `pip install "oriented-det[viewer]"`
-- For **ONNX export**: `uv pip install -e ".[export]"` then `python -m export --help` (see [ONNX export](docs/examples/export.md))
+- For **ONNX export**: `uv pip install -e ".[export]"` then `odet export --help` (see [ONNX export](docs/examples/export.md))
 - For **macOS Apple Silicon** or **CPU-only**, see [Installation](docs/getting-started/installation.md) and the [macOS walkthrough](https://deeplearning.earth/posts/2026-06-25_oriented_object_detection_on_macos_in_pure_python/) (`odet image-demo` on MPS, no CUDA toolchain).
 - Verify: `pytest tests/test_geometry.py tests/test_iou.py tests/test_nms.py`
 
@@ -94,7 +94,7 @@ After that, copy-pasted commands and unmodified configs that reference `/path/to
 | **Configs** | [`configs/`](configs/) | Experiment JSON (`_base_` inheritance, schema in `configs/config.schema.json`) |
 | **Runs** | `runs/<model_type>/<timestamp>/` | Checkpoints, `config.json` snapshot, `train.log` (created at train time; not shipped in the repo) |
 | **Docs** | [`docs/`](docs/) | MkDocs user guide and API reference |
-| **Examples** | [`demo/`](demo/), [`pretrained/`](pretrained/), [`notebooks/`](notebooks/), [`export/`](export/) | Demo images; Hub checkpoints; Kaggle FAIR1M tutorial notebook; ONNX export producer (`python -m export`) |
+| **Examples** | [`demo/`](demo/), [`pretrained/`](pretrained/), [`notebooks/`](notebooks/), [`export/`](export/) | Demo images; Hub checkpoints; Kaggle FAIR1M tutorial notebook; ONNX export producer (`odet export`) |
 
 **`odet` vs `tools/`:** Installing the package registers the `odet` command. It loads modules under `tools/` (for example `tools.train`, `tools.save_predictions`). Shared inference and collate code lives in [`oriented_det/runtime/`](oriented_det/runtime/). You do not need two workflows — use **`odet`** (or **`make`**, which calls `odet`).
 
@@ -105,7 +105,7 @@ After that, copy-pasted commands and unmodified configs that reference `/path/to
 Full documentation is in the **docs/** folder and can be built and served with MkDocs:
 
 - **Build/serve**: `make docs` or `make docs-serve` (see [docs/README.md](docs/README.md)); or `uv pip install -e ".[docs]"` then `mkdocs serve`.
-- **Guides**: [Getting Started](docs/getting-started/installation.md), [User Guide](docs/user-guide/geometry.md), [API Reference](docs/api/geometry.md), [Examples](docs/examples/inference.md), [ONNX export](docs/examples/export.md).
+- **Guides**: [Getting Started](docs/getting-started/installation.md), [User Guide](docs/user-guide/geometry.md), [API Reference](docs/api/geometry.md), [Examples](docs/examples/inference.md), [Docker deploy](docs/examples/deploy.md), [ONNX export](docs/examples/export.md).
 - **Blog**: [DeepLearning.Earth](https://deeplearning.earth) — curated posts below (code, metrics, inference knobs).
 
 | Topic | Post |
@@ -121,6 +121,8 @@ Full documentation is in the **docs/** folder and can be built and served with M
 | FCOS vs Oriented R-CNN (MPS latency, thresholds) | [https://deeplearning.earth/posts/2026-09-02_rotated_fcos_vs_oriented_rcnn_on_macos/](https://deeplearning.earth/posts/2026-09-02_rotated_fcos_vs_oriented_rcnn_on_macos/) |
 | Side-by-side browser demo (three 1× checkpoints) | [https://deeplearning.earth/posts/2026-09-06_oriented_det_optical_satellite_demo/](https://deeplearning.earth/posts/2026-09-06_oriented_det_optical_satellite_demo/) |
 | Apache 2.0 vs DOTA / HRSC dataset terms | [https://deeplearning.earth/posts/2026-09-10_oriented_det_apache_license_versus_dota/](https://deeplearning.earth/posts/2026-09-10_oriented_det_apache_license_versus_dota/) |
+| Docker deploy (Tile Geo Process) | [https://deeplearning.earth/posts/2026-10-05_deploy_oriented_det_in_docker/](https://deeplearning.earth/posts/2026-10-05_deploy_oriented_det_in_docker/) |
+| ONNX export (no PyTorch on infer) | [https://deeplearning.earth/posts/2026-10-08_onnx_export_without_pytorch/](https://deeplearning.earth/posts/2026-10-08_onnx_export_without_pytorch/) |
 
 ## Documentation by folder
 
@@ -131,8 +133,8 @@ Full documentation is in the **docs/** folder and can be built and served with M
 | [oriented_det/cli/](oriented_det/cli/) | [oriented_det/cli/README.md](oriented_det/cli/README.md) | **`odet`** entry point and subcommand list |
 | [tools/](tools/) | [tools/README.md](tools/README.md) | CLI script implementations (invoked by `odet`; see [Repository layout](#repository-layout)) |
 | [configs/](configs/) | [configs/README.md](configs/README.md) | DOTA configs and **pretrain model zoo** (`_base_` inheritance) |
-| [deploy/example/](deploy/example/) | [deploy/example/README.md](deploy/example/README.md) | Minimal DOTA deploy smoke image |
-| [export/](export/) | [export/README.md](export/README.md) | ONNX export producer (`python -m export`); consumer bundle in gitignored `onnx_export/` |
+| [deploy/example/](deploy/example/) | [deploy/example/README.md](deploy/example/README.md) | Sanic Tile Geo Process Docker image (DOTA smoke) |
+| [export/](export/) | [export/README.md](export/README.md) | ONNX export producer (`odet export`); consumer bundle in gitignored `onnx_export/` |
 | [docs/](docs/) | [docs/README.md](docs/README.md) | MkDocs source; full user guide and API reference |
 
 ## Training and evaluation
@@ -149,7 +151,7 @@ Install once: `uv pip install -e .`. Then:
 | COCO polygons → DOTA | `odet coco-to-dota --ann-file instances.json --image-dir images --output-dir out` |
 | Val predictions | `odet preds --experiment-dir runs/oriented_rcnn/<timestamp>` or `make preds` |
 | Offline mAP | `make eval-val` or `make preds` then `make metrics`. DOTA: leaky val tiles (real test is Task 1). HRSC/FAIR1M/SSDD/HRSID: held-out test/val. |
-| ONNX export | `make export-onnx` or `python -m export onnx ...` (see [export/README.md](export/README.md)) |
+| ONNX export | `odet export onnx ...` or `make export-onnx` (see [export/README.md](export/README.md)) |
 
 DOTA configs: per-model `dota_le90_1x.json` under [configs/](configs/) (3× where published; Faster R-CNN and FCOS DOTA are 1×). Run `odet --help` for all subcommands. Makefile shortcuts and script-level options: [tools/README.md](tools/README.md). Config reference: [docs/user-guide/configuration.md](docs/user-guide/configuration.md), [configs/config.schema.json](configs/config.schema.json), [configs/README.md](configs/README.md).
 
@@ -185,7 +187,7 @@ OrientedDet is Apache-2.0 open source. For **consulting, training workshops, or
 
 ## Publishing to PyPI
 
-Version **0.3.0** — tag releases as **`v0.3.0`** (git) matching `version` in `pyproject.toml`.
+Version **0.3.1** — tag releases as **`v0.3.1`** (git) matching `version` in `pyproject.toml`.
 
 Configs: edit **`configs/`** at the repo root, then **`make sync-configs`** so **`oriented_det/configs/`** stays in sync (see **`oriented_det/configs/vendored_manifest.txt`**). CI runs **`make check-configs`**.
 
@@ -216,7 +218,7 @@ make check-configs
 pytest tests/ -q
 ```
 
-Bump version in `pyproject.toml` and `docs/changelog.md` if needed (currently **0.3.0**).
+Bump version in `pyproject.toml` and `docs/changelog.md` if needed (currently **0.3.1**).
 
 #### 3. Build locally
 

{oriented_det-0.3.0 → oriented_det-0.3.1}/export/__init__.py

@@ -1,4 +1,4 @@
-"""ONNX export package for OrientedDet. CLI: ``python -m export``.
+"""ONNX export package for OrientedDet. CLI: ``odet export`` / ``python -m export``.
 
 Inference (``demo`` / ``infer``) needs only numpy, Pillow, and onnxruntime —
 see ``requirements-runtime.txt``. The same consumer modules are copied into

{oriented_det-0.3.0 → oriented_det-0.3.1}/export/__main__.py

@@ -1,4 +1,4 @@
-"""``python -m export`` entry point."""
+"""``python -m export`` entry point (same as ``odet export``)."""
 
 from export.cli import main
 

{oriented_det-0.3.0 → oriented_det-0.3.1}/export/cli.py

@@ -1,4 +1,4 @@
-"""Standalone CLI for ONNX export (``python -m export``)."""
+"""ONNX export CLI (``odet export`` / ``python -m export``)."""
 
 from __future__ import annotations
 
@@ -15,7 +15,7 @@ _COMMANDS: Dict[str, Tuple[str, str]] = {
     "preds": ("export.scripts.save_predictions_onnx", "export-preds"),
 }
 
-# Former odet / export-tf names. Export is not an odet subcommand.
+# Legacy python -m export names. Top-level odet export-* (TF / SavedModel) stay removed.
 _ALIASES = {
     "export-onnx": "onnx",
     "export-infer": "infer",
@@ -30,14 +30,15 @@ def install_hint() -> str:
     export = root / "requirements-export.txt"
     return (
         f"Infer/demo (no oriented-det): pip install -r {runtime}\n"
-        f"Export from a checkpoint: pip install -r {export} (plus oriented-det)"
+        f"Producer (odet export): uv pip install -e \".[export]\" "
+        f"or pip install -r {export} (plus oriented-det)"
     )
 
 
 def _print_help() -> None:
-    print("Usage: python -m export <command> [options]")
-    print("Run from the oriented-det repo root. Not an odet subcommand.")
-    print("demo / infer do not need oriented-det.")
+    print("Usage: odet export <command> [options]")
+    print("       python -m export <command> [options]")
+    print("ONNX only (no TensorFlow / SavedModel). demo / infer do not need PyTorch.")
     print("")
     print("Commands:")
     for name in sorted(_COMMANDS):
@@ -45,14 +46,14 @@ def _print_help() -> None:
     print("")
     print("Examples:")
     print(
-        "  python -m export onnx --config path/to/config.json "
+        "  odet export onnx --config path/to/config.json "
         "--checkpoint path/to/model.pth --output ./onnx_export/model.onnx"
     )
     print(
-        "  python -m export infer --onnx ./onnx_export/model.onnx "
+        "  odet export infer --onnx ./onnx_export/model.onnx "
         "--images ./tiles --output ./onnx_export/predictions"
     )
-    print("  python -m export demo")
+    print("  odet export demo")
     print("")
     print(install_hint())
 

{oriented_det-0.3.0 → oriented_det-0.3.1}/export/nms.py

@@ -1,6 +1,6 @@
 """CPU rotated IoU and greedy NMS.
 
-Source of truth: ``export/nms.py``. ``python -m export onnx`` copies this file
+Source of truth: ``export/nms.py``. ``odet export onnx`` copies this file
 with the rest of the consumer stack next to ``model.onnx``. No oriented-det imports.
 
 IoU backends:

oriented_det-0.3.1/export/scripts/__init__.py

@@ -0,0 +1,6 @@
+"""CLI implementations invoked by ``odet export`` / ``python -m export``."""
+
+from __future__ import annotations
+
+__all__ = []
+

{oriented_det-0.3.0 → oriented_det-0.3.1}/export/scripts/demo.py

@@ -1,8 +1,8 @@
 #!/usr/bin/env python3
 """Run the exported FCOS ONNX on the bundled plane image and assert NMS.
 
-Works from the repo (``python -m export demo``) and from a copied
-``onnx_export/`` bundle (``python demo.py``). No oriented-det.
+Works from the repo (``odet export demo`` / ``python -m export demo``) and from a copied
+``onnx_export/`` bundle (``python demo.py``). The copied bundle does not need oriented-det.
 """
 
 from __future__ import annotations

{oriented_det-0.3.0 → oriented_det-0.3.1}/export/scripts/infer_onnx.py

@@ -1,8 +1,8 @@
 #!/usr/bin/env python3
 """Run a pre-NMS ONNX export on a folder of images (ORT + Python NMS).
 
-Works from the repo (``python -m export infer``) and from a copied
-``onnx_export/`` bundle (``python infer_onnx.py ...``). No oriented-det.
+Works from the repo (``odet export infer`` / ``python -m export infer``) and from a copied
+``onnx_export/`` bundle (``python infer_onnx.py ...``). The copied bundle does not need oriented-det.
 """
 
 from __future__ import annotations

{oriented_det-0.3.0 → oriented_det-0.3.1}/oriented_det/cli/__init__.py

@@ -28,6 +28,7 @@ _COMMANDS: Dict[str, Tuple[str, str]] = {
     "labels-to-comma": ("tools.dota_labels_to_comma", "odet-labels-to-comma"),
     "free-gpu": ("tools.free_gpu", "odet-free-gpu"),
     "pretrained": ("tools.pretrained_download", "odet-pretrained"),
+    "export": ("export.cli", "odet-export"),
 }
 
 
@@ -42,6 +43,7 @@ def _print_help() -> None:
     print("  odet train --config configs/oriented_rcnn/dota_le90_1x.json")
     print("  odet train-multi-gpu --config configs/oriented_rcnn/dota_le90_1x.json")
     print("  odet preds --experiment-dir runs/oriented_rcnn/<id>")
+    print("  odet export onnx --config path/to/config.json --checkpoint path/to/model.pth")
     print("  odet playground-csv --data-root /path/to/playground")
 
 

{oriented_det-0.3.0 → oriented_det-0.3.1}/oriented_det/configs/rotated_retinanet/dota_le90_1x.json

@@ -104,7 +104,7 @@
     ],
     "retinanet_stacked_convs": 4,
     "box_reg_loss_type": "l1",
-    "use_hbb_for_matching": true,
+    "use_hbb_for_matching": false,
     "final_nms_use_cpu": false,
     "inference_pre_nms_score_threshold": 0.05,
     "final_nms_iou_threshold": 0.1,
@@ -113,7 +113,7 @@
     "rpn_negative_iou_threshold": 0.4
   },
   "production": {
-    "score_threshold": 0.35,
+    "score_threshold": 0.25,
     "inference_pre_nms_score_threshold": 0.05,
     "final_nms_iou_threshold": 0.1,
     "max_detections_per_image": 3000,

oriented_det-0.3.1/oriented_det/configs/rotated_retinanet/dota_le90_1x_hbb.json

@@ -0,0 +1,9 @@
+{
+  "_base_": ["./dota_le90_1x.json"],
+  "model": {
+    "use_hbb_for_matching": true
+  },
+  "production": {
+    "score_threshold": 0.35
+  }
+}

oriented_det-0.3.1/oriented_det/configs/rotated_retinanet/dota_le90_3x_hbb.json

@@ -0,0 +1,7 @@
+{
+  "_base_": ["./dota_le90_1x_hbb.json"],
+  "training": {
+    "num_epochs": 36,
+    "lr_scheduler_milestones": [24, 33]
+  }
+}

{oriented_det-0.3.0 → oriented_det-0.3.1}/oriented_det/models/rotated_retinanet.py

@@ -377,6 +377,8 @@ def compute_oriented_retinanet_loss(
         - "loss_classifier": Classification loss (sigmoid focal loss, normalized by the number
           of positive anchors); same TensorBoard name as two-stage detectors
         - "loss_box_reg": Box regression loss (smooth L1)
+        - "retinanet_num_pos" / "retinanet_num_neg" / "retinanet_num_ignore": assignment
+          counts for the batch (Python floats, not part of ``total_loss``)
     """
     if torch is None or F is None:
         raise RuntimeError("PyTorch is required for loss computation.")
@@ -403,6 +405,8 @@ def compute_oriented_retinanet_loss(
     # normalized once by the total number of positive anchors in the batch (avg_factor).
     cls_loss_sums = []
     num_pos_total = 0
+    num_neg_total = 0
+    num_ignore_total = 0
     reg_loss_sums: List[torch.Tensor] = []
     reg_loss_decoded_addends: List[torch.Tensor] = []
     num_reg_pos_total = 0
@@ -585,7 +589,9 @@ def compute_oriented_retinanet_loss(
                     )
                 )
             num_pos_total += int((labels == 1).sum())
-            
+            num_neg_total += int((labels == 0).sum())
+            num_ignore_total += int((labels == -1).sum())
+
             positive_mask = labels == 1
             if not positive_mask.any():
                 continue
@@ -731,6 +737,9 @@ def compute_oriented_retinanet_loss(
     return {
         "loss_classifier": loss_classification,
         "loss_box_reg": loss_box_reg,
+        "retinanet_num_pos": float(num_pos_total),
+        "retinanet_num_neg": float(num_neg_total),
+        "retinanet_num_ignore": float(num_ignore_total),
     }
 
 

{oriented_det-0.3.0 → oriented_det-0.3.1}/oriented_det/ops/diff_iou_rotated.py

@@ -9,8 +9,8 @@ Runs on CPU or CUDA (vectorized PyTorch; no custom kernel). Intersection
 math is fp32 with AMP off, same reason as KFIoU.
 
 Vertex order around the intersection centroid uses ``atan2`` (mmcv dropped
-its CUDA sort for stability). Invalid slots gather a zeroed dummy so extra
-shoelace terms are 0.
+its CUDA sort for stability). Unused hull slots repeat the first intersection
+vertex so extra shoelace terms are 0.
 """
 
 from __future__ import annotations
@@ -110,7 +110,12 @@ def _box1_in_box2(corners1: Tensor, corners2: Tensor) -> Tensor:
 
 
 def _sort_indices(vertices: Tensor, mask: Tensor) -> Tensor:
-    """``(B, N, 24, 2)`` + validity → ``(B, N, 9)`` hull indices (wrap + dummy pad)."""
+    """``(B, N, 24, 2)`` + validity → ``(B, N, 9)`` hull indices (wrap + pad).
+
+    Unused slots repeat the first hull vertex so extra shoelace terms are 0.
+    Padding with an invalid exterior corner (still has real xy) inflates the
+    polygon and clamps IoU to 1.
+    """
     mask_b = mask.bool()
     mask_f = mask_b.to(dtype=vertices.dtype)
     num_valid = mask_f.sum(dim=-1, keepdim=True).clamp(min=1.0)
@@ -121,17 +126,12 @@ def _sort_indices(vertices: Tensor, mask: Tensor) -> Tensor:
     sorted_idx = torch.argsort(ang, dim=-1)
     first8 = sorted_idx[..., :8]
     wrap = first8[..., :1]
-    dummy_idx = torch.argsort(mask_b.to(dtype=vertices.dtype), dim=-1)[..., :1]
     k = mask_b.to(dtype=torch.int64).sum(dim=-1).clamp(max=8)
     slots = torch.arange(9, device=vertices.device).view(1, 1, 9)
-    k_exp = k.unsqueeze(-1)
-    is_vertex = slots < k_exp
-    is_wrap = (slots == k_exp) & (k_exp > 0)
-    first8_pad = torch.cat([first8, first8[..., :1]], dim=-1)
-    dummy_exp = dummy_idx.expand(-1, -1, 9)
+    is_vertex = slots < k.unsqueeze(-1)
+    first8_pad = torch.cat([first8, wrap], dim=-1)
     wrap_exp = wrap.expand(-1, -1, 9)
-    out = torch.where(is_vertex, first8_pad, dummy_exp)
-    return torch.where(is_wrap, wrap_exp, out)
+    return torch.where(is_vertex, first8_pad, wrap_exp)
 
 
 def _intersection_area(corners1: Tensor, corners2: Tensor) -> Tensor:

{oriented_det-0.3.0 → oriented_det-0.3.1}/oriented_det/pretrained/hub.py

@@ -207,7 +207,7 @@ def download_asset(
     """Download a manifest asset from Hugging Face Hub.
 
     Args:
-        asset_name: Manifest slug (e.g. ``rotated_retinanet_dota_le90_3x``) or ``.pth`` filename.
+        asset_name: Manifest slug (e.g. ``rotated_retinanet_dota_le90_1x``) or ``.pth`` filename.
         dest: Optional explicit destination file. Defaults to ``<pretrained_dir>/<filename>``.
 
     Returns:

{oriented_det-0.3.0 → oriented_det-0.3.1}/oriented_det/pretrained/manifest.json

@@ -3,22 +3,40 @@
   "revision": "main",
   "assets": {
     "rotated_retinanet_dota_le90_1x": {
-      "filename": "rotated_retinanet_r50_fpn_dota_le90_1x-9eb38d49.pth",
-      "sha256": "9eb38d49819413f26fa32273b5954966f834caddf0f4d6dfde569864e0b68139",
+      "filename": "rotated_retinanet_r50_fpn_dota_le90_1x-b9190270.pth",
+      "sha256": "b9190270d771b6e5ae2041444b1f2ab5dc3963b09447d3352c76fcef2c72d0cc",
       "source_recipe": "configs/rotated_retinanet/dota_le90_1x.json",
+      "experiment_dir": "runs/rotated_retinanet/20260919-093746",
+      "eval_map50": 0.7112,
+      "eval_task1_map50": 0.7172,
+      "eval_report": "docs/eval-reports/rotated_retinanet_dota_le90_1x/model_analysis.md"
+    },
+    "rotated_retinanet_dota_le90_1x_hbb": {
+      "filename": "rotated_retinanet_r50_fpn_dota_le90_1x_hbb-9eb38d49.pth",
+      "sha256": "9eb38d49819413f26fa32273b5954966f834caddf0f4d6dfde569864e0b68139",
+      "source_recipe": "configs/rotated_retinanet/dota_le90_1x_hbb.json",
       "experiment_dir": "runs/rotated_retinanet/20260912-105343",
       "eval_map50": 0.6820,
       "eval_task1_map50": 0.6787,
-      "eval_report": "docs/eval-reports/rotated_retinanet_dota_le90_1x/model_analysis.md"
+      "eval_report": "docs/eval-reports/rotated_retinanet_dota_le90_1x_hbb/model_analysis.md"
     },
     "rotated_retinanet_dota_le90_3x": {
-      "filename": "rotated_retinanet_r50_fpn_dota_le90_3x-42968545.pth",
-      "sha256": "429685453dc18b03d7ec659643fc3419745a361bd8d947a0cb98a88bb562ba9a",
+      "filename": "rotated_retinanet_r50_fpn_dota_le90_3x-961aaf73.pth",
+      "sha256": "961aaf7371be236418fe9f90d2223b18a03109cdedf40a242d093d12ad118c8d",
       "source_recipe": "configs/rotated_retinanet/dota_le90_3x.json",
+      "experiment_dir": "runs/rotated_retinanet/20260920-054001",
+      "eval_map50": 0.7856,
+      "eval_task1_map50": 0.7389,
+      "eval_report": "docs/eval-reports/rotated_retinanet_dota_le90_3x/model_analysis.md"
+    },
+    "rotated_retinanet_dota_le90_3x_hbb": {
+      "filename": "rotated_retinanet_r50_fpn_dota_le90_3x_hbb-42968545.pth",
+      "sha256": "429685453dc18b03d7ec659643fc3419745a361bd8d947a0cb98a88bb562ba9a",
+      "source_recipe": "configs/rotated_retinanet/dota_le90_3x_hbb.json",
       "experiment_dir": "runs/rotated_retinanet/20260913-031811",
       "eval_map50": 0.7651,
-      "eval_task1_map50": 0.7070,
-      "eval_report": "docs/eval-reports/rotated_retinanet_dota_le90_3x/model_analysis.md"
+      "eval_task1_map50": 0.707,
+      "eval_report": "docs/eval-reports/rotated_retinanet_dota_le90_3x_hbb/model_analysis.md"
     },
     "rotated_faster_rcnn_dota_le90_3x": {
       "filename": "rotated_faster_rcnn_r50_fpn_dota_le90_3x-9951acc6.pth",

{oriented_det-0.3.0 → oriented_det-0.3.1}/oriented_det/train/engine.py

@@ -847,6 +847,14 @@ def train_one_epoch(
                 f"match_rate={avg_metrics.get('roi_match_rate', 0.0):.2%}",
                 flush=True,
             )
+        if "retinanet_num_pos" in avg_metrics:
+            print(
+                "  RetinaNet assign: "
+                f"pos={avg_metrics.get('retinanet_num_pos', 0.0):.1f}, "
+                f"neg={avg_metrics.get('retinanet_num_neg', 0.0):.1f}, "
+                f"ignore={avg_metrics.get('retinanet_num_ignore', 0.0):.1f}",
+                flush=True,
+            )
         # Debug: full loss breakdown to compare with MMRotate and spot unbalanced losses.
         if debug:
             loss_keys = [k for k in avg_metrics if k not in ("epoch", "_global_step") and (k == "total_loss" or k.startswith("loss_"))]

{oriented_det-0.3.0 → oriented_det-0.3.1}/oriented_det.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: oriented-det
-Version: 0.3.0
+Version: 0.3.1
 Summary: A lightweight, modern PyTorch library for rotated object detection in aerial and satellite imagery
 Author: Jeff Faudi, DL4EO
 Maintainer: Jeff Faudi, DL4EO
@@ -77,7 +77,7 @@ Dynamic: license-file
 - **Datasets**: DOTA polygon loader (pattern, split file, or separate folders), **HRSC2016** native XML loader, **FAIR1M** 37-class XML loader (+ convert/tile tutorial), **SSDD** and **HRSID** SAR ship loaders (finetune DOTA 1× Hub locally; no SAR zoo), image tiling, label filtering, ignore masks, oriented mAP evaluation
 - **Models**: **Oriented R-CNN** ([Xie et al., ICCV 2021](https://openaccess.thecvf.com/content/ICCV2021/html/Xie_Oriented_R-CNN_for_Object_Detection_ICCV_2021_paper.html); horizontal RPN + MidpointOffset → oriented RoIAlign + oriented ROI head), **Rotated Faster R-CNN** (Ren et al., NeurIPS 2015 two-stage baseline with horizontal RPN + horizontal RoIAlign + rotated ROI head; MMRotate reference), **Rotated RetinaNet** ([Lin et al., ICCV 2017](https://openaccess.thecvf.com/content_ICCV_2017/papers/Lin_Focal_Loss_for_ICCV_2017_paper.pdf); oriented anchors, sigmoid focal loss), **Rotated FCOS** (anchor-free single-stage; distance-angle coder, centerness, L1 / KFIoU / decoded rIoU); ResNet + FPN backbones; selective loading of external checkpoints where configs wire `checkpoint.load_from_checkpoint`
 - **Training**: JSON configs + **`odet train`**, mixed precision (AMP), gradient accumulation, checkpointing, best-metric tracking, TensorBoard, optional curriculum learning and profiling
-- **ONNX export**: pre-NMS ONNX + Python NMS (`python -m export` / `make export-onnx`); consumers run ONNX Runtime only
+- **ONNX export**: pre-NMS ONNX + Python NMS (`odet export` / `make export-onnx`); consumers run ONNX Runtime only
 
 Hands-on write-ups (code, metrics, inference knobs) live on **[DeepLearning.Earth](https://deeplearning.earth)** — especially the [15 DOTA class tour](https://deeplearning.earth/posts/2026-06-23_oriented_rcnn_detections_for_the_15_dota_classes/), [sampled rIoU and ProbIoU](https://deeplearning.earth/posts/2026-07-10_rotated_faster_rcnn_probiou_dota/), [Rotated FCOS / v0.2.0](https://deeplearning.earth/posts/2026-08-28_oriented-det_v0_2_0_rotated_fcos_decoded_riou_and_the_updated_zoo/), and [macOS FCOS vs Oriented R-CNN](https://deeplearning.earth/posts/2026-09-02_rotated_fcos_vs_oriented_rcnn_on_macos/).
 
@@ -107,7 +107,7 @@ uv pip install -e .
 - From PyPI: `pip install oriented-det`
 - For development and tests: `uv pip install -e ".[dev]"`
 - For the Gradio prediction viewer: `uv pip install -e ".[viewer]"` or `pip install "oriented-det[viewer]"`
-- For **ONNX export**: `uv pip install -e ".[export]"` then `python -m export --help` (see [ONNX export](docs/examples/export.md))
+- For **ONNX export**: `uv pip install -e ".[export]"` then `odet export --help` (see [ONNX export](docs/examples/export.md))
 - For **macOS Apple Silicon** or **CPU-only**, see [Installation](docs/getting-started/installation.md) and the [macOS walkthrough](https://deeplearning.earth/posts/2026-06-25_oriented_object_detection_on_macos_in_pure_python/) (`odet image-demo` on MPS, no CUDA toolchain).
 - Verify: `pytest tests/test_geometry.py tests/test_iou.py tests/test_nms.py`
 
@@ -154,7 +154,7 @@ After that, copy-pasted commands and unmodified configs that reference `/path/to
 | **Configs** | [`configs/`](configs/) | Experiment JSON (`_base_` inheritance, schema in `configs/config.schema.json`) |
 | **Runs** | `runs/<model_type>/<timestamp>/` | Checkpoints, `config.json` snapshot, `train.log` (created at train time; not shipped in the repo) |
 | **Docs** | [`docs/`](docs/) | MkDocs user guide and API reference |
-| **Examples** | [`demo/`](demo/), [`pretrained/`](pretrained/), [`notebooks/`](notebooks/), [`export/`](export/) | Demo images; Hub checkpoints; Kaggle FAIR1M tutorial notebook; ONNX export producer (`python -m export`) |
+| **Examples** | [`demo/`](demo/), [`pretrained/`](pretrained/), [`notebooks/`](notebooks/), [`export/`](export/) | Demo images; Hub checkpoints; Kaggle FAIR1M tutorial notebook; ONNX export producer (`odet export`) |
 
 **`odet` vs `tools/`:** Installing the package registers the `odet` command. It loads modules under `tools/` (for example `tools.train`, `tools.save_predictions`). Shared inference and collate code lives in [`oriented_det/runtime/`](oriented_det/runtime/). You do not need two workflows — use **`odet`** (or **`make`**, which calls `odet`).
 
@@ -165,7 +165,7 @@ After that, copy-pasted commands and unmodified configs that reference `/path/to
 Full documentation is in the **docs/** folder and can be built and served with MkDocs:
 
 - **Build/serve**: `make docs` or `make docs-serve` (see [docs/README.md](docs/README.md)); or `uv pip install -e ".[docs]"` then `mkdocs serve`.
-- **Guides**: [Getting Started](docs/getting-started/installation.md), [User Guide](docs/user-guide/geometry.md), [API Reference](docs/api/geometry.md), [Examples](docs/examples/inference.md), [ONNX export](docs/examples/export.md).
+- **Guides**: [Getting Started](docs/getting-started/installation.md), [User Guide](docs/user-guide/geometry.md), [API Reference](docs/api/geometry.md), [Examples](docs/examples/inference.md), [Docker deploy](docs/examples/deploy.md), [ONNX export](docs/examples/export.md).
 - **Blog**: [DeepLearning.Earth](https://deeplearning.earth) — curated posts below (code, metrics, inference knobs).
 
 | Topic | Post |
@@ -181,6 +181,8 @@ Full documentation is in the **docs/** folder and can be built and served with M
 | FCOS vs Oriented R-CNN (MPS latency, thresholds) | [https://deeplearning.earth/posts/2026-09-02_rotated_fcos_vs_oriented_rcnn_on_macos/](https://deeplearning.earth/posts/2026-09-02_rotated_fcos_vs_oriented_rcnn_on_macos/) |
 | Side-by-side browser demo (three 1× checkpoints) | [https://deeplearning.earth/posts/2026-09-06_oriented_det_optical_satellite_demo/](https://deeplearning.earth/posts/2026-09-06_oriented_det_optical_satellite_demo/) |
 | Apache 2.0 vs DOTA / HRSC dataset terms | [https://deeplearning.earth/posts/2026-09-10_oriented_det_apache_license_versus_dota/](https://deeplearning.earth/posts/2026-09-10_oriented_det_apache_license_versus_dota/) |
+| Docker deploy (Tile Geo Process) | [https://deeplearning.earth/posts/2026-10-05_deploy_oriented_det_in_docker/](https://deeplearning.earth/posts/2026-10-05_deploy_oriented_det_in_docker/) |
+| ONNX export (no PyTorch on infer) | [https://deeplearning.earth/posts/2026-10-08_onnx_export_without_pytorch/](https://deeplearning.earth/posts/2026-10-08_onnx_export_without_pytorch/) |
 
 ## Documentation by folder
 
@@ -191,8 +193,8 @@ Full documentation is in the **docs/** folder and can be built and served with M
 | [oriented_det/cli/](oriented_det/cli/) | [oriented_det/cli/README.md](oriented_det/cli/README.md) | **`odet`** entry point and subcommand list |
 | [tools/](tools/) | [tools/README.md](tools/README.md) | CLI script implementations (invoked by `odet`; see [Repository layout](#repository-layout)) |
 | [configs/](configs/) | [configs/README.md](configs/README.md) | DOTA configs and **pretrain model zoo** (`_base_` inheritance) |
-| [deploy/example/](deploy/example/) | [deploy/example/README.md](deploy/example/README.md) | Minimal DOTA deploy smoke image |
-| [export/](export/) | [export/README.md](export/README.md) | ONNX export producer (`python -m export`); consumer bundle in gitignored `onnx_export/` |
+| [deploy/example/](deploy/example/) | [deploy/example/README.md](deploy/example/README.md) | Sanic Tile Geo Process Docker image (DOTA smoke) |
+| [export/](export/) | [export/README.md](export/README.md) | ONNX export producer (`odet export`); consumer bundle in gitignored `onnx_export/` |
 | [docs/](docs/) | [docs/README.md](docs/README.md) | MkDocs source; full user guide and API reference |
 
 ## Training and evaluation
@@ -209,7 +211,7 @@ Install once: `uv pip install -e .`. Then:
 | COCO polygons → DOTA | `odet coco-to-dota --ann-file instances.json --image-dir images --output-dir out` |
 | Val predictions | `odet preds --experiment-dir runs/oriented_rcnn/<timestamp>` or `make preds` |
 | Offline mAP | `make eval-val` or `make preds` then `make metrics`. DOTA: leaky val tiles (real test is Task 1). HRSC/FAIR1M/SSDD/HRSID: held-out test/val. |
-| ONNX export | `make export-onnx` or `python -m export onnx ...` (see [export/README.md](export/README.md)) |
+| ONNX export | `odet export onnx ...` or `make export-onnx` (see [export/README.md](export/README.md)) |
 
 DOTA configs: per-model `dota_le90_1x.json` under [configs/](configs/) (3× where published; Faster R-CNN and FCOS DOTA are 1×). Run `odet --help` for all subcommands. Makefile shortcuts and script-level options: [tools/README.md](tools/README.md). Config reference: [docs/user-guide/configuration.md](docs/user-guide/configuration.md), [configs/config.schema.json](configs/config.schema.json), [configs/README.md](configs/README.md).
 
@@ -245,7 +247,7 @@ OrientedDet is Apache-2.0 open source. For **consulting, training workshops, or
 
 ## Publishing to PyPI
 
-Version **0.3.0** — tag releases as **`v0.3.0`** (git) matching `version` in `pyproject.toml`.
+Version **0.3.1** — tag releases as **`v0.3.1`** (git) matching `version` in `pyproject.toml`.
 
 Configs: edit **`configs/`** at the repo root, then **`make sync-configs`** so **`oriented_det/configs/`** stays in sync (see **`oriented_det/configs/vendored_manifest.txt`**). CI runs **`make check-configs`**.
 
@@ -276,7 +278,7 @@ make check-configs
 pytest tests/ -q
 ```
 
-Bump version in `pyproject.toml` and `docs/changelog.md` if needed (currently **0.3.0**).
+Bump version in `pyproject.toml` and `docs/changelog.md` if needed (currently **0.3.1**).
 
 #### 3. Build locally
 

{oriented_det-0.3.0 → oriented_det-0.3.1}/oriented_det.egg-info/SOURCES.txt

@@ -62,7 +62,9 @@ oriented_det/configs/rotated_fcos/hrsc2016_le90_1x.json
 oriented_det/configs/rotated_fcos/hrsc2016_le90_3x.json
 oriented_det/configs/rotated_fcos/hrsid_le90_1x.json
 oriented_det/configs/rotated_retinanet/dota_le90_1x.json
+oriented_det/configs/rotated_retinanet/dota_le90_1x_hbb.json
 oriented_det/configs/rotated_retinanet/dota_le90_3x.json
+oriented_det/configs/rotated_retinanet/dota_le90_3x_hbb.json
 oriented_det/data/__init__.py
 oriented_det/data/airbus_playground.py
 oriented_det/data/build.py