def start_offline_dashboard(sid, filepath):
num_files = len(os.listdir(os.path.join(filepath, "rgb")))
sio.emit("handleMaxFrames",
num_files - 1) # Minus 1 because filenames start at 00000
def retrain_detector(sid, settings={}):
inference_json = LP.retrain_detector(settings)
sio.emit("annotationRetrain", inference_json)
def update_log_settings(sid, new_values):
self.log_settings["image_resolution"] = new_values["image_resolution"]
self.log_settings["image_quality"] = new_values["image_quality"]
sio.emit("image_settings", self.log_settings)
def send_chat(self, chat):
"""Send chat from agent to player"""
logging.info("Sending chat: {}".format(chat))
sio.emit("showAssistantReply", {"agent_reply": "Agent: {}".format(chat)})
self.memory.add_chat(self.memory.self_memid, chat)
return self.cagent.send_chat(chat)
def log(self, rgb_depth, detections, humans, old_rgb_depth):
"""Log all relevant data from the perceptual models for the dashboard.
All the data here is sent to the dashboard using a socketio emit event. This data
is then used by the dashboard for different debugging visualizations. Add any data
that you would want to fetch on the dashboard here.
Args:
rgb_depth (RGBDepth): the current RGBDepth frame. This frame might be different from
old_rgb_depth, which is the RGBDepth frame for which SlowPerception has been run
detections (list[Detections]): list of all detections
humans (list[Human]): list of all humans detected
old_rgb_depth (RGBDepth): RGBDepth frame for which detections and humans are being sent.
This is the frame for which SlowPerception has been run.
"""
if hasattr(sio, "mock"):
return
sio.emit("image_settings", self.log_settings)
resolution = self.log_settings["image_resolution"]
quality = self.log_settings["image_quality"]
serialized_image = rgb_depth.to_struct(resolution, quality)
if old_rgb_depth is not None:
serialized_object_image = old_rgb_depth.to_struct(resolution, quality)
else:
serialized_object_image = -1
height, width = rgb_depth.rgb.shape[0], rgb_depth.rgb.shape[1]
scale = float(resolution) / height
new_height, new_width = resolution, int(resolution * float(width) / height)
serialized_objects = [x.to_struct() for x in detections] if detections is not None else []
serialized_humans = [x.to_struct() for x in humans] if humans is not None else []
sio.emit("rgb", serialized_image["rgb"])
sio.emit(
"depth",
{
"depthImg": serialized_image["depth_img"],
"depthMax": serialized_image["depth_max"],
"depthMin": serialized_image["depth_min"],
},
)
sio.emit(
"objects",
{
"image": serialized_object_image,
"objects": serialized_objects,
"height": new_height,
"width": new_width,
"scale": scale,
},
)
sio.emit(
"humans",
{
"image": serialized_object_image,
"humans": serialized_humans,
"height": new_height,
"width": new_width,
"scale": scale,
},
)