OpenGait/.sisyphus/notepads/demo-visualizer/learnings.md

## Task 1: CLI Flag Addition

- Added --visualize boolean flag to opengait/demo/__main__.py
- Uses argparse with action="store_true" as requested
- Passes visualize=args.visualize to ScoliosisPipeline constructor
- Note: ScoliosisPipeline does not yet accept visualize parameter (Task 3 will add this)
- All existing CLI options preserved with same defaults as pipeline.py click definitions


## Task 1 Final State

- Single if __name__ == "__main__" block
- Uses inspect.signature to conditionally pass visualize
- All CLI options preserved with correct defaults
- --visualize flag present and functional


## Task 2: OpenCVVisualizer Implementation

### Completed
- Created opengait/demo/visualizer.py with OpenCVVisualizer class
- Internal state self.mask_mode = 0 (0: Both, 1: Raw, 2: Normalized)
- Method update(frame, bbox, track_id, mask_raw, silhouette, label, confidence, fps)
- Two windows: main stream (bbox + text overlay) and segmentation (mode-dependent view)
- Key handling: m cycles mask mode, q returns False to signal quit
- close() method calls cv2.destroyAllWindows()

### Key Implementation Details
- Silhouette shape assumed 64x44 (from preprocess.py SIL_HEIGHT/SIL_WIDTH)
- Upscaled to 256x176 for display using INTER_NEAREST to preserve pixelation
- Handles missing inputs gracefully (bbox/mask_raw/silhouette can be None)
- Converts grayscale arrays to BGR for consistent display
- Mode indicator text shown in segmentation window for operator clarity

### Type Checking Notes
- OpenCV MatLike type conflicts with NDArray annotations
- These are acceptable warnings - runtime behavior is correct
- Used _ = cv2.function() pattern to suppress unused return value warnings


## Task 2 Retry: Type Annotation Fixes

### Problem
LSP diagnostics showed basedpyright errors due to OpenCV MatLike type conflicting with strict NDArray[np.uint8] annotations.

### Solution
- Added ImageArray type alias = NDArray[np.uint8]
- Used typing.cast() to convert OpenCV return values (MatLike) to ImageArray
- Applied cast() on all cv2.cvtColor() and cv2.resize() calls that return MatLike

### Changes Made
- Added: from typing import cast
- Added: ImageArray = NDArray[np.uint8] type alias
- Modified: All methods returning NDArray[np.uint8] now use cast() for OpenCV calls
- Modified: Parameter types changed from NDArray[np.uint8] to ImageArray for consistency

### Verification
- lsp_diagnostics: 0 errors (4 warnings about Any types remain - acceptable)
- Import test: uv run python -c "from opengait.demo.visualizer import OpenCVVisualizer; print(ok)" -> ok
- No TODO/FIXME/HACK placeholders introduced


## Task 3: Pipeline Integration - COMPLETED

### Implementation Summary
- Added `_visualizer: object | None` attribute to `ScoliosisPipeline` class
- Updated `__init__` to accept `visualize: bool = False` parameter
- Conditionally instantiates `OpenCVVisualizer` when `visualize=True`
- Updated `_select_silhouette` to return 4-tuple: `(silhouette, mask_raw, bbox, track_id)`
- Updated `process_frame` to:
  - Unpack 4-tuple from `_select_silhouette`
  - Return visualization payload dict with `mask_raw`, `bbox`, `silhouette`, `track_id`, `label`, `confidence`
  - Return payload in all paths (preprocess-only, not-ready-to-classify, and classified)
- Updated `run()` to:
  - Compute per-frame EMA FPS with alpha=0.1 smoothing
  - Call `self._visualizer.update()` with all required parameters
  - Break loop if visualizer returns `False` (user pressed 'q')
- Updated `close()` to close visualizer via `self._visualizer.close()` in finally path

### Key Design Decisions
- Used `object | None` type for `_visualizer` to avoid circular import issues
- EMA FPS uses alpha=0.1 for reasonable smoothing while maintaining responsiveness
- Visualization payload returned in all paths ensures consistent behavior
- Visualizer cleanup happens in `close()` which is called in `finally` block

### Verification Results
- `uv run python -m opengait.demo --help` - PASSED
- `uv run python -m opengait.demo --source foo --checkpoint bar --config baz --device cpu --visualize` - PASSED (reaches file-not-found as expected)
- Constructor accepts `visualize` parameter - PASSED

## Task 3 Regression Fix - COMPLETED

### Problem
LSP errors at lines 343-350 due to calling `.get()` / `.update()` on `object` typed values (`viz_payload`, `_visualizer`).

### Solution
Used `cast()` to tell the type checker the actual types:
1. `viz_payload` is cast to `dict[str, object]` before calling `.get()`
2. `self._visualizer` is cast to `object` and methods are accessed via `getattr()`

### Key Changes
- In `run()`: `viz_dict = cast(dict[str, object], viz_payload)` before `.get()` calls
- In `run()`: `visualizer = cast(object, self._visualizer)` then `getattr(visualizer, "update", None)`
- In `close()`: Same pattern for calling `.close()` on visualizer

### Verification
- `lsp_diagnostics(opengait/demo/pipeline.py)` - ZERO ERRORS (only acceptable warnings)
- `uv run python -m opengait.demo --help` - PASSED
- `uv run python -m opengait.demo --source foo --checkpoint bar --config baz --device cpu --visualize` - PASSED (reaches file-not-found)



## Task 4: CLI Validation Pattern

### Lesson
When migrating from click to argparse, ensure ALL behavior is preserved:
- Input validation (both CLI-level and runtime)
- Exit code mapping for different error types
- User-friendly error messages

### Pattern for CLI Entry Points
```python
if __name__ == "__main__":
    args = parser.parse_args()
    
    # CLI-level validation
    if args.preprocess_only and not args.silhouette_export_path:
        print("Error: ...", file=sys.stderr)
        raise SystemExit(2)
    
    try:
        validate_runtime_inputs(...)
        pipeline = ScoliosisPipeline(...)
        raise SystemExit(pipeline.run())
    except ValueError as err:
        print(f"Error: {err}", file=sys.stderr)
        raise SystemExit(2) from err
    except RuntimeError as err:
        print(f"Runtime error: {err}", file=sys.stderr)
        raise SystemExit(1) from err
```

### Key Takeaway
Exit code parity matters for test suites and shell scripting. ValueError (user input errors) -> 2, RuntimeError (system/runtime errors) -> 1.

## Task 5: YOLO Model Path Relocation

### Task Summary
Moved yolo11n-seg.pt from repo root to ckpt/yolo11n-seg.pt and updated all in-repo references.

### Files Modified
1. opengait/demo/__main__.py - Updated --yolo-model default
2. opengait/demo/pipeline.py - Updated --yolo-model default
3. tests/demo/test_pipeline.py - Updated YOLO_MODEL_PATH

### Verification
- All 3 references now point to ckpt/yolo11n-seg.pt
- Tests pass: 12 passed, 1 skipped in 37.35s



## Oracle Caveat Fix #1: Simplified inspect-based conditional logic

### Change Summary
- Removed `import inspect` from opengait/demo/__main__.py
- Removed `inspect.signature(ScoliosisPipeline.__init__)` call
- Removed conditional `if "visualize" in sig.parameters:` check
- Now passes `visualize=args.visualize` directly in pipeline_kwargs

### Rationale
The inspect-based conditional was unnecessary complexity for an intra-package API contract.
ScoliosisPipeline.__init__ explicitly accepts the visualize parameter, so the runtime
check provided no value and added maintenance burden.

### Verification
- `uv run python -m opengait.demo --help` - PASSED (all CLI options preserved)
- `uv run pytest tests/demo/test_pipeline.py -q` - PASSED (12 passed, 1 skipped)
- `lsp_diagnostics(opengait/demo/__main__.py)` - No new errors (only pre-existing warnings)

### Behavior Parity
CLI behavior unchanged. --visualize flag still supported and passed to pipeline.
Default yolo path preserved: ckpt/yolo11n-seg.pt


## Oracle Caveat Fix #2: Explicit Typing for Visualizer

### Change Summary
- Replaced loose `_visualizer: object | None` with explicit `OpenCVVisualizer | None` using TYPE_CHECKING forward reference
- Removed `cast(object, ...)` and `getattr(..., "update"/"close")` indirection in favor of direct method calls
- Added explicit casts for values extracted from viz_payload dict to satisfy type checker

### Rationale
The previous implementation used `object` typing and runtime introspection (`getattr`) to avoid circular imports and maintain optional dependency semantics. However, this sacrificed type safety and code clarity. Using `TYPE_CHECKING` allows us to:
1. Import the actual type for static analysis without runtime import
2. Call methods directly on the typed visualizer instance
3. Maintain lazy import behavior (OpenCVVisualizer still only imported when visualize=True)

### Key Changes
1. Added TYPE_CHECKING block with forward import: `from .visualizer import OpenCVVisualizer`
2. Changed `_visualizer` type annotation from `object | None` to `OpenCVVisualizer | None`
3. In `run()`: Removed `cast(object, self._visualizer)` and `getattr(visualizer, "update", None)` pattern
   - Now calls `self._visualizer.update(...)` directly
   - Added explicit casts for dict values extracted from viz_payload
4. In `close()`: Removed `cast(object, self._visualizer)` and `getattr(visualizer, "close", None)` pattern
   - Now calls `self._visualizer.close()` directly

### Verification
- `lsp_diagnostics(opengait/demo/pipeline.py)` - ZERO ERRORS (only pre-existing warnings unrelated to visualizer)
- `uv run pytest tests/demo/test_pipeline.py -q` - PASSED (12 passed, 1 skipped)
- Runtime behavior unchanged: lazy import preserved, EMA FPS calculation unchanged, quit handling unchanged

### Type Safety Improvement
Before: Runtime introspection required, no static type checking on visualizer methods
After: Full static type checking on visualizer.update() and visualizer.close() calls, proper type inference for all parameters

## Oracle Non-blocking Improvement: _prepare_both_view Redundant Work Removal

### Change Summary
- Modified `_prepare_both_view` in `opengait/demo/visualizer.py` to eliminate wasted text-rendering work
- Previously: Called `_prepare_raw_view` and `_prepare_normalized_view` which drew mode indicators, then converted to grayscale (destroying the text), then drew combined indicator
- Now: Inlines the view preparation logic without mode indicators, preserving only the final combined indicator

### Rationale
The sub-view mode indicators ("Raw Mask" and "Normalized") were being drawn and immediately destroyed by grayscale conversion before stacking. This was pure overhead with no visual effect. The final combined indicator ("Both: Raw | Normalized") is the only one visible to users.

### Behavior Preservation
- Visual output unchanged: Final combined mode indicator still displayed
- Mode toggle semantics untouched: mask_mode cycling (0->1->2->0) unchanged
- Placeholder handling preserved: None inputs still produce zero-filled arrays
- All existing tests pass: 12 passed, 1 skipped

### Code Quality Impact
- Reduced unnecessary OpenCV text rendering operations
- Eliminated redundant BGR->Gray->BGR conversions on sub-views
- Improved maintainability by making the wasted work explicit (removed)

### Verification
- `lsp_diagnostics(opengait/demo/visualizer.py)` - 0 errors (4 pre-existing Any warnings)
- `uv run pytest tests/demo/test_pipeline.py -q` - PASSED (12 passed, 1 skipped)


## Oracle Non-blocking Cleanup: Duplicate Import Removal

### Change Summary
- Removed duplicated import block at lines 17-28 in `tests/demo/test_pipeline.py`
- Duplicated imports: `json`, `pickle`, `Path`, `subprocess`, `sys`, `time`, `Final`, `cast`, `pytest`, `torch`, `ScoNetDemo`
- Kept first import block (lines 1-15) intact

### Rationale
Identical import block appeared twice consecutively—likely from a merge conflict resolution or copy-paste error. No functional impact, but code hygiene improvement.

### Verification
- `uv run pytest tests/demo/test_pipeline.py -q` - PASSED (12 passed, 1 skipped)
- `lsp_diagnostics(tests/demo/test_pipeline.py)` - No new errors (only pre-existing warnings)

### Behavior Preservation
- No test logic modified
- All imports required by tests remain available
- Import order and formatting unchanged


## Dependency Configuration: cvmmap-client Local Path Source

### Task Summary
Added `cvmmap-client` as a dependency sourced from local path `/home/crosstyan/Code/cvmmap-python-client`.

### Changes Made
1. **pyproject.toml**:
   - Added `"cvmmap-client"` to `[project] dependencies list
   - Added `[tool.uv.sources]` section with path mapping:
     ```toml
     [tool.uv.sources]
     cvmmap-client = { path = "/home/crosstyan/Code/cvmmap-python-client" }
     ```

2. **uv.lock**: Updated automatically via `uv lock` to include:
   - `cvmmap-client v0.1.0` (from file:// path)
   - `pyzmq v27.1.0` (transitive dependency)

### Verification
- `uv lock` - PASSED (resolved 104 packages)
- `uv run python -c "from cvmmap import CvMmapClient; print('ok')"` - PASSED

### Key Points
- Package name in provider repo: `cvmmap-client` (distribution name)
- Import path: `from cvmmap import CvMmapClient` (module name)
- uv path sources require absolute paths
- Lockfile captures the path dependency for reproducibility

## Bugfix: _FrameMetadata NameError in cvmmap_source()

### Root Cause
The `_FrameMetadata` Protocol class was defined inside a `TYPE_CHECKING` block, making it available only during type checking but not at runtime. However, `cvmmap_source()` used `_FrameMetadata` in a runtime function annotation:

```python
async def _async_generator() -> AsyncIterator[tuple[np.ndarray, _FrameMetadata]]:
```

This caused `NameError: name '_FrameMetadata' is not defined` when the function was called.

### Fix
Moved both `_FrameMetadata` and `_CvMmapClient` Protocol classes out of the `TYPE_CHECKING` block to module level. These protocols are lightweight (just method signatures) and don't require heavy imports, so there's minimal runtime cost.

### Verification
- `lsp_diagnostics`: Only 1 warning (unnecessary type: ignore comment)
- `uv run python -c "from opengait.demo.input import cvmmap_source"`: Import successful
- `uv run pytest tests/demo/`: 74 passed, 3 skipped
- Runtime test confirms generator creation works without NameError

### Pattern to Remember
When using Protocol classes for type annotations:
- If the annotation is used in a nested function (defined inside another function), it needs to be available at runtime
- TYPE_CHECKING-guarded types can only be used in annotations that are evaluated at type-check time
- Module-level function annotations are evaluated at definition time (runtime)

## Fix: Window Freeze on No-Detection Frames

### Problem
When YOLO doesn't detect any person in a frame, `process_frame()` returns `None`. The visualizer update was conditional on `viz_payload is not None`, causing the OpenCV window to freeze when no detections were present for extended periods.

### Solution
Modified `ScoliosisPipeline.run()` in `opengait/demo/pipeline.py` to always call `self._visualizer.update()` when visualization is enabled, even when `viz_payload` is `None`. When no detection data is available, the visualizer receives `None` values for bbox, mask_raw, silhouette, label, and confidence, which it handles gracefully by displaying placeholder views.

### Changes Made
1. **opengait/demo/pipeline.py** (lines 344-393):
   - Changed condition from `if self._visualizer is not None and viz_payload is not None:` to `if self._visualizer is not None:`
   - Added else branch to set default values (all None/0) when `viz_payload is None`
   - Visualizer now receives frame and FPS data every frame, preventing window freeze

2. **tests/demo/test_pipeline.py**:
   - Added `MockVisualizer` class to track update calls
   - Added `test_pipeline_visualizer_updates_on_no_detection()` regression test
   - Test verifies visualizer is called for all frames even when YOLO returns no detections

### Verification
- New regression test passes (RED -> GREEN TDD cycle)
- All 13 existing pipeline tests pass (1 skipped)
- No new LSP errors introduced
- Visualizer handles None values gracefully (existing behavior)

## Task 1: Stabilize overlapped duplicate code blocks - COMPLETED

### Changes Made to opengait/demo/pipeline.py

1. **Fixed tuple unpacking in _select_silhouette (lines 181-187)**:
   - Changed `mask_raw, bbox, track_id = selected` to `mask_raw, bbox_mask, bbox_frame, track_id = selected`
   - Updated `mask_to_silhouette()` call to use `bbox_mask` (mask coordinate space)
   - Changed return statement to use `bbox_frame` (frame coordinate space)
   - This fixes the 4-tuple unpacking from `select_person()` which returns `(mask, bbox_mask, bbox_frame, track_id)`

2. **Added _last_viz_payload attribute**:
   - Added `_last_viz_payload: dict[str, object] | None` class field after `_visualizer`
   - Initialized to `None` in `__init__` after visualizer initialization

3. **Updated visualizer logic in run() for no-detection caching**:
   - Changed condition from `if self._visualizer is not None and viz_payload is not None:` to just `if self._visualizer is not None:`
   - Added caching logic: when `viz_payload is not None`, cache a copy with `v.copy() if hasattr(v, 'copy') else v` for each value
   - Use cached payload when current frame has no detection: `viz_data = viz_payload if viz_payload is not None else self._last_viz_payload`
   - Handle case when no cache exists: use default None/0 values
   - Visualizer now updates every frame when enabled, preventing window freeze

### Verification Results
- LSP diagnostics: No new critical errors (only pre-existing warnings)
- Tests: 14 passed, 1 skipped
- Import test: OK

### Key Implementation Detail
The caching logic uses a dict comprehension with `.copy()` for mutable values (numpy arrays) to prevent mutation issues, satisfying the test requirement that cached masks should be copies, not the same object reference.


## Task: BBoxXYXY Type Alias Introduction - COMPLETED

### Summary
Introduced and applied `BBoxXYXY` alias for bbox semantics across demo module files.
The alias was already defined in `preprocess.py`; this task applied it consistently to
`pipeline.py` which had raw `tuple[int, int, int, int]` annotations.

### Changes Made
1. **opengait/demo/pipeline.py**:
   - Added import: `from .preprocess import BBoxXYXY, frame_to_person_mask, mask_to_silhouette`
   - Changed `_select_silhouette` return type annotation (line 176):
     - Before: `tuple[int, int, int, int]`
     - After: `BBoxXYXY`
   - Changed fallback cast annotation (line 192):
     - Before: `tuple[UInt8[ndarray, "h w"], tuple[int, int, int, int]] | None`
     - After: `tuple[UInt8[ndarray, "h w"], BBoxXYXY] | None`
   - Changed bbox cast in run() method (line 372):
     - Before: `cast(tuple[int, int, int, int] | None, bbox_obj)`
     - After: `cast(BBoxXYXY | None, bbox_obj)`

### Files Already Using BBoxXYXY (No Changes Needed)
- **opengait/demo/window.py**: Already imports and uses `BBoxXYXY` for `select_person()` return type
- **opengait/demo/visualizer.py**: Already imports and uses `BBoxXYXY` for `_draw_bbox()` and `update()` parameters
- **opengait/demo/preprocess.py**: Defines `BBoxXYXY = tuple[int, int, int, int]` with docstring

### Verification
- `grep 'tuple\[int, int, int, int\]'` in demo/ scope: Only matches alias definition (preprocess.py line 27)
- `lsp_diagnostics`: No new errors introduced from changes
- `uv run pytest tests/demo/test_pipeline.py -q`: 14 passed, 1 skipped

### Semantic Preservation
- No runtime behavior changes - purely type annotation cleanup
- Mask-space vs frame-space bbox semantics unchanged
- All bbox coordinate calculations preserved exactly

## 2026-02-28: TypedDict Contract Applied for DemoResult

### Changes Made
1. **opengait/demo/output.py**: `create_result()` already returns `DemoResult` TypedDict (confirmed)
2. **opengait/demo/pipeline.py**: 
   - Added `DemoResult` to imports from `.output`
   - Changed `_result_buffer` type from `list[dict[str, object]]` to `list[DemoResult]`
   - Fixed pre-existing type error in visualization payload caching by extracting dict comprehension to explicit loop with `Callable[[], object]` cast

### Key Patterns
- TypedDict provides structural typing for dictionary results
- `DemoResult` fields: `frame`, `track_id`, `label`, `confidence`, `window`, `timestamp_ns`
- Pipeline buffer now enforces consistent result typing throughout the pipeline

### Type Safety
- `output.py`: No diagnostics (clean)
- `pipeline.py`: Only warnings (no errors) - warnings are pre-existing pyarrow/dynamic import related
- All 14 tests pass

## 2026-02-28: Visualization Regression Fixes - BBox Scale and Both-Mode Raw Mask

### Issue A: BBox Scale Bug in Fallback Path

**Problem:** In `_select_silhouette()` fallback path (when `select_person()` returns None), the bbox from `frame_to_person_mask()` was in mask-space coordinates but was being drawn on the full-size frame, causing the overlay to appear much smaller than actual.

**Root Cause:** 
- `frame_to_person_mask()` returns bbox in mask coordinate space (scaled down from original frame)
- Primary path through `select_person()` correctly returns both `bbox_mask` and `bbox_frame`
- Fallback path was returning mask-space bbox directly without conversion

**Fix:** 
- Added frame-space bbox conversion in fallback path using `result.orig_shape`
- Calculates scale factors: `scale_x = frame_w / mask_w`, `scale_y = frame_h / mask_h`
- Applies scaling to bbox coordinates: `bbox_frame = (int(bbox_mask[0] * scale_x), ...)`
- Safely falls back to mask-space bbox if `orig_shape` unavailable or dimensions invalid

**Code Location:** `opengait/demo/pipeline.py` lines 206-226 in `_select_silhouette()`

### Issue B: Both-Mode Raw Mask Appears Empty

**Problem:** In `_prepare_both_view()`, the raw mask pane appeared empty/black in "Both" display mode.

**Root Cause:**
- Primary path returns mask as float32 with values in [0, 1] range
- Fallback path returns mask as uint8 with values in [0, 255] range
- OpenCV display expects uint8 [0, 255] for proper visualization
- Float32 [0, 1] values displayed as nearly black

**Fix:**
- Added dtype normalization in `_prepare_both_view()` before resizing
- Check: `if mask_gray.dtype == np.float32 or mask_gray.dtype == np.float64:`
- Convert: `mask_gray = (mask_gray * 255).astype(np.uint8)`
- Ensures consistent uint8 [0, 255] display format regardless of input dtype

**Code Location:** `opengait/demo/visualizer.py` lines 330-332 in `_prepare_both_view()`

### Verification
- All 14 pipeline tests pass (1 skipped)
- LSP diagnostics show only pre-existing warnings
- No new tests needed - existing test coverage sufficient



## 2026-02-28: Tuple-to-TypedDict Migration for select_person and _select_silhouette

### Summary
Successfully migrated two function return contracts from anonymous tuples to explicit TypedDict structures:
1. `window.select_person()` → `PersonSelection | None`
2. `pipeline._select_silhouette()` → `SilhouetteSelection | None`

### Key TypedDict Definitions

**PersonSelection** (in `opengait/demo/window.py`):
```python
class PersonSelection(TypedDict):
    mask: "Float[ndarray, 'h w']"  # jaxtyping annotated
    bbox_mask: "tuple[int, int, int, int]"
    bbox_frame: "tuple[int, int, int, int]"
    track_id: int
```

**SilhouetteSelection** (in `opengait/demo/pipeline.py`):
```python
class SilhouetteSelection(TypedDict):
    silhouette: "Float[ndarray, '64 44']"  # jaxtyping annotated
    mask_raw: "UInt8[ndarray, 'h w']"     # jaxtyping annotated
    bbox: "BBoxXYXY"
    track_id: int
```

### Migration Pattern
- Replace `tuple[X, Y, Z] | None` return type with `TypedDictName | None`
- Replace tuple unpacking at call sites with dict key access
- Update all return statements to return dict literals
- Ensure jaxtyping annotations are preserved on ndarray fields

### Files Modified
- `opengait/demo/window.py` - Added PersonSelection TypedDict, updated select_person
- `opengait/demo/pipeline.py` - Added SilhouetteSelection TypedDict, updated _select_silhouette
- `tests/demo/test_window.py` - Updated assertions to use dict keys
- `tests/demo/test_pipeline.py` - Updated mock return to use dict

### Verification
- `uv run pytest tests/demo/test_window.py -q` → 22 passed, 1 skipped
- `uv run pytest tests/demo/test_pipeline.py -q` → 14 passed, 1 skipped
- No errors in lsp_diagnostics (only pre-existing warnings)

### Behavior Confirmation
No behavior change - purely a typing/contract refactor. All visualization, caching, and bbox scaling logic remains unchanged.


## 2026-02-28: ProcessFrameResult TypedDict Added

### Summary
Added `ProcessFrameResult` TypedDict for `ScoliosisPipeline.process_frame()` return contract.

### ProcessFrameResult Definition (pipeline.py)
```python
class ProcessFrameResult(TypedDict):
    """Structured return type for process_frame method.

    Contains visualization payload and optional classification result.
    Uses jaxtyping for array fields to preserve shape information.
    """

    mask_raw: "UInt8[ndarray, 'h w'] | None"
    bbox: "BBoxXYXY | None"
    silhouette: "Float[ndarray, '64 44'] | None"
    track_id: int
    label: "str | None"
    confidence: "float | None"
    result: "DemoResult | None"
```

### Files Modified
- `opengait/demo/pipeline.py` - Added ProcessFrameResult TypedDict, updated process_frame return type

### Changes Made
1. Added `ProcessFrameResult` TypedDict with jaxtyping annotations for array fields
2. Updated `process_frame() -> ProcessFrameResult | None`
3. Updated all return statements to include `result` field (None for non-classified paths)
4. Updated `_last_viz_payload` type from `dict[str, object]` to `ProcessFrameResult`
5. Updated cast in `run()` from `dict[str, object]` to `ProcessFrameResult`

### Verification
- `uv run pytest tests/demo/test_pipeline.py -q` → 14 passed, 1 skipped
- No errors in lsp_diagnostics (only pre-existing warnings)

### Behavior Confirmation
No behavior change - purely a typing/contract refactor.


## 2026-02-28: Type Quality Cleanup (Oracle Feedback)

### Changes Made

1. **window.py - PersonSelection bbox type consistency**
   - Changed `bbox_mask` and `bbox_frame` types from `"tuple[int, int, int, int]"` to `"BBoxXYXY"`
   - Uses the existing type alias for semantic clarity and consistency

2. **pipeline.py - run() visualization payload cleanup**
   - Replaced generic `cached: dict[str, object]` with `cached: ProcessFrameResult`
   - Replaced `.get()` access with direct key access on `ProcessFrameResult`
   - Simplified extraction: removed intermediate `*_obj` variables and casting cascade
   - Maintained copy semantics for mutable numpy arrays (mask_raw, silhouette)

### Benefits
- Type-safe caching with explicit ProcessFrameResult structure
- Direct key access eliminates unnecessary `.get()` calls and type casts
- Consistent use of BBoxXYXY alias across window and pipeline modules

### Verification
- `uv run pytest tests/demo/test_window.py -q` → 22 passed, 1 skipped
- `uv run pytest tests/demo/test_pipeline.py -q` → 14 passed, 1 skipped
- No runtime behavior changes


## 2026-02-28: bbox_frame Documentation Added

### Documentation Changes

Updated docstrings to clarify coordinate space semantics for bbox fields:

**PersonSelection (window.py)**:
```
Coordinate spaces:
    - bbox_mask: Bounding box in mask coordinate space (XYXY: x1, y1, x2, y2).
      This is the downsampled mask resolution (e.g., 100x100), used for
      silhouette extraction and mask-space operations.
    - bbox_frame: Bounding box in original frame pixel coordinates (XYXY: x1, y1, x2, y2).
      This is the full-resolution frame size (e.g., 640x480), used for
      visualization overlay on the original video frame.
```

**SilhouetteSelection (pipeline.py)**:
```
The bbox field is in original frame pixel coordinates (XYXY: x1, y1, x2, y2),
suitable for visualization overlay on the original video frame.
```

**ProcessFrameResult (pipeline.py)**:
```
The bbox field is in original frame pixel coordinates (XYXY: x1, y1, x2, y2),
suitable for visualization overlay on the original video frame.
```

**select_person function (window.py)**:
```
- bbox_mask: bounding box in mask coordinate space (XYXY format: x1, y1, x2, y2).
  This is the downsampled mask resolution, used for silhouette extraction.
- bbox_frame: bounding box in original frame pixel coordinates (XYXY format: x1, y1, x2, y2).
  This is the full-resolution frame size, used for visualization overlay.
```

### Key Distinction
- `bbox_mask`: Mask-space coordinates (downsampled, for processing)
- `bbox_frame`: Frame-space coordinates (full resolution, for visualization)

### Verification
- `uv run pytest tests/demo/test_window.py tests/demo/test_pipeline.py -q` → 36 passed, 1 skipped


## 2026-02-28: bbox_frame Documentation in select_person

### Docstring Updated (window.py)

Updated `select_person` function docstring to explicitly document coordinate spaces:

```python
Returns:
    PersonSelection dictionary with keys:
    - mask: the person's segmentation mask
    - bbox_mask: bounding box in mask coordinate space (XYXY format: x1, y1, x2, y2).
      This is the downsampled mask resolution (e.g., 100x100), used for
      silhouette extraction and mask-space operations.
    - bbox_frame: bounding box in original frame pixel coordinates (XYXY format: x1, y1, x2, y2).
      This is the full-resolution frame size (e.g., 640x480), used for
      visualization overlay on the original video frame.
    - track_id: the person's track ID
```

### Key Distinction
- `bbox_mask`: Mask-space coordinates (downsampled resolution, for processing)
- `bbox_frame`: Frame-space coordinates (full resolution, for visualization)

### Verification
- `uv run pytest tests/demo/test_window.py -q` → 22 passed


## 2026-02-28: pkl Alias for silhouette-export-format

### Summary
Added `pkl` as an alias for `pickle` in `--silhouette-export-format` CLI option.

### Implementation
In `opengait/demo/pipeline.py` `__init__` method:
```python
self._silhouette_export_format = silhouette_export_format
# Normalize format alias: pkl -> pickle
if self._silhouette_export_format == "pkl":
    self._silhouette_export_format = "pickle"
```

### Behavior
- `pkl` → normalized to `pickle` → pickle export
- `pickle` → unchanged → pickle export
- `parquet` → unchanged → parquet export
- `invalid` → unchanged → ValueError on export

### Verification
- All existing tests pass
- Custom validation test confirms:
  - pkl alias works
  - pickle still works
  - parquet still works
  - invalid formats still rejected