adding detect_markers: can now detect and report camera pose w.r.t on…

…e or multiple planes, and also individual markers

requirements.txt

@@ -1,2 +1,2 @@
-glassesTools[GUI]==1.3.4    # pulls in various other dependencies such as opencv, imgui_bundle, numpy, pandas
+glassesTools[GUI]==1.4.0    # pulls in various other dependencies such as opencv, imgui_bundle, numpy, pandas
 glassesValidator==2.0.0

src/gazeMapper/process/detect_markers.py

@@ -0,0 +1,78 @@
+import pathlib
+import threading
+
+from glassesTools import aruco, marker, plane as gt_plane
+from glassesTools.video_gui import GUI, generic_tooltip_drawer, qns_tooltip
+
+
+from . import naming
+from .. import config, episode, plane, session
+
+
+stopAllProcessing = False
+def process(working_dir: str|pathlib.Path, config_dir: str|pathlib.Path, rec_type: session.RecordingType, show_visualization=False, show_rejected_markers=False):
+    # if show_visualization, each frame is shown in a viewer, overlaid with info about detected markers and poster
+    # if show_rejected_markers, rejected ArUco marker candidates are also shown in the viewer. Possibly useful for debug
+    working_dir = pathlib.Path(working_dir)
+    config_dir  = pathlib.Path(config_dir)
+
+    print('processing: {}'.format(working_dir.name))
+
+    # if we need gui, we run processing in a separate thread (GUI needs to be on the main thread for OSX, see https://github.com/pthom/hello_imgui/issues/33)
+    if show_visualization:
+        gui = GUI(use_thread = False)
+        gui.set_interesting_keys('qns')
+        gui.register_draw_callback('status',lambda: generic_tooltip_drawer(qns_tooltip()))
+        gui.add_window(working_dir.name)
+
+        proc_thread = threading.Thread(target=do_the_work, args=(working_dir, config_dir, gui, show_rejected_markers, rec_type))
+        proc_thread.start()
+        gui.start()
+        proc_thread.join()
+        return stopAllProcessing
+    else:
+        return do_the_work(working_dir, config_dir, None, False, rec_type)
+
+
+def do_the_work(working_dir: pathlib.Path, config_dir: pathlib.Path, gui: GUI, show_rejected_markers: bool, rec_type: session.RecordingType):
+    global stopAllProcessing
+
+    # get info about recording
+    in_video = session.read_recording_info(working_dir, rec_type)[1]
+
+    # get info about the study it is a part of
+    study_config = config.Study.load_from_json(config_dir)
+
+    # get interval(s) coded to be analyzed, if any
+    episodes = episode.list_to_marker_dict(episode.read_list_from_file(working_dir / 'coding.tsv'))
+
+    # process the above into a dict of plane definitions and a dict with frame number intervals for which to use each
+    planes = {v for k in study_config.planes_per_interval for v in study_config.planes_per_interval[k]}
+    planes_setup = {}
+    analyze_frames = {}
+    for p in planes:
+        p_def = [pl for pl in study_config.planes if pl.name==p][0]
+        pl = plane.get_plane_from_definition(p_def, config_dir/p)
+        planes_setup[p] = {'plane': pl, 'aruco_dict': p_def.aruco_dict, 'aruco_params': {'markerBorderBits': p_def.marker_border_bits}, 'min_num_markers': p_def.min_num_markers}
+        # determine for which frames this plane should be used
+        anal_episodes = [k for k in study_config.planes_per_interval if p in study_config.planes_per_interval[k]]
+        all_episodes = [ep for k in anal_episodes for ep in episodes[k]]
+        analyze_frames[p] = sorted(all_episodes, key = lambda x: x[1])
+
+    stopAllProcessing, poses, individual_markers = \
+        aruco.run_pose_estimation(in_video, working_dir / "frameTimestamps.tsv", working_dir / "calibration.xml",   # input video
+                                  # output
+                                  working_dir,
+                                  # intervals to process
+                                  analyze_frames,
+                                  # detector and pose estimator setup
+                                  planes_setup, config.get_marker_dict_from_list(study_config.individual_markers),
+                                  # visualization setup
+                                  gui, 8, show_rejected_markers)
+
+    for p in poses:
+        gt_plane.write_list_to_file(poses[p], working_dir/f'{naming.plane_pose_prefix}{p}.tsv', skip_failed=True)
+    for i in individual_markers:
+        marker.write_list_to_file(individual_markers[i], working_dir/f'{naming.marker_pose_prefix}{i}.tsv', skip_failed=True)
+
+    return stopAllProcessing

src/gazeMapper/process/naming.py

@@ -0,0 +1,4 @@
+
+plane_pose_prefix   = 'planePose_'
+marker_pose_prefix  = 'markerPose_'
+world_gaze_prefix   = 'planeGaze_'