feat: face recognition agent #53

Open
2584433 wants to merge 31 commits from feat/face-recognition into main
4 changed files with 62 additions and 33 deletions
Showing only changes of commit 985327de70 - Show all commits

View File

@@ -8,7 +8,7 @@ from zmq.asyncio import Context
from control_backend.agents import BaseAgent from control_backend.agents import BaseAgent
from control_backend.agents.actuation.robot_gesture_agent import RobotGestureAgent from control_backend.agents.actuation.robot_gesture_agent import RobotGestureAgent
from control_backend.agents.perception.visual_emotion_recognition_agent.visual_emotion_recognition_agent import ( from control_backend.agents.perception.visual_emotion_recognition_agent.visual_emotion_recognition_agent import ( # noqa
VisualEmotionRecognitionAgent, VisualEmotionRecognitionAgent,
) )
from control_backend.core.config import settings from control_backend.core.config import settings

View File

@@ -4,28 +4,50 @@ from collections import Counter, defaultdict
import cv2 import cv2
import numpy as np import numpy as np
from pydantic_core import ValidationError
import zmq import zmq
import zmq.asyncio as azmq import zmq.asyncio as azmq
from control_backend.agents.perception.visual_emotion_recognition_agentvisual_emotion_recognizer import ( # noqa
from control_backend.agents import BaseAgent
from control_backend.agents.perception.visual_emotion_recognition_agent.visual_emotion_recognizer import (
DeepFaceEmotionRecognizer, DeepFaceEmotionRecognizer,
) )
from pydantic_core import ValidationError
from control_backend.agents import BaseAgent
from control_backend.core.agent_system import InternalMessage from control_backend.core.agent_system import InternalMessage
from control_backend.core.config import settings from control_backend.core.config import settings
from control_backend.schemas.belief_message import Belief from control_backend.schemas.belief_message import Belief
# START FROM RI COMMUNICATION AGENT?
class VisualEmotionRecognitionAgent(BaseAgent): class VisualEmotionRecognitionAgent(BaseAgent):
def __init__(self, name, socket_address: str, bind: bool = False, timeout_ms: int = 1000): def __init__(self, name: str, socket_address: str, bind: bool = False, timeout_ms: int = 1000,
window_duration:
int = settings.behaviour_settings.visual_emotion_recognition_window_duration_s
, min_frames_required: int =
settings.behaviour_settings.visual_emotion_recognition_min_frames_per_face):
"""
Initialize the Visual Emotion Recognition Agent.
:param name: Name of the agent
:param socket_address: Address of the socket to connect or bind to
:param bind: Whether to bind to the socket address (True) or connect (False)
:param timeout_ms: Timeout for socket receive operations in milliseconds
:param window_duration: Duration in seconds over which to aggregate emotions
:param min_frames_required: Minimum number of frames per face required to consider a face
valid
"""
super().__init__(name) super().__init__(name)
self.socket_address = socket_address self.socket_address = socket_address
self.socket_bind = bind self.socket_bind = bind
self.timeout_ms = timeout_ms self.timeout_ms = timeout_ms
self.window_duration = window_duration
self.min_frames_required = min_frames_required
async def setup(self): async def setup(self):
"""
Initialize the agent resources.
1. Initializes the :class:`VisualEmotionRecognizer`.
2. Connects to the video input ZMQ socket.
3. Starts the background emotion recognition loop.
"""
self.logger.info("Setting up %s.", self.name) self.logger.info("Setting up %s.", self.name)
self.emotion_recognizer = DeepFaceEmotionRecognizer() self.emotion_recognizer = DeepFaceEmotionRecognizer()
@@ -45,17 +67,16 @@ class VisualEmotionRecognitionAgent(BaseAgent):
async def emotion_update_loop(self): async def emotion_update_loop(self):
""" """
Retrieve a video frame from the input socket. Background loop to receive video frames, recognize emotions, and update beliefs.
1. Receives video frames from the ZMQ socket.
2. Uses the :class:`VisualEmotionRecognizer` to detect emotions.
3. Aggregates emotions over a time window.
4. Sends updates to the BDI Core Agent about detected emotions.
""" """
window_duration = 5 # seconds # Next time to process the window and update emotions
next_window_time = time.time() + window_duration next_window_time = time.time() + self.window_duration
# To detect false positives
# Minimal number of frames a face has to be detected to consider it valid
# Can also reduce false positives by ignoring faces that are too small; not implemented
# Also use face confidence thresholding in recognizer
min_frames_required = 2
# Tracks counts of detected emotions per face index
face_stats = defaultdict(Counter) face_stats = defaultdict(Counter)
prev_dominant_emotions = set() prev_dominant_emotions = set()
@@ -82,20 +103,19 @@ class VisualEmotionRecognitionAgent(BaseAgent):
# If window duration has passed, process the collected stats # If window duration has passed, process the collected stats
if time.time() >= next_window_time: if time.time() >= next_window_time:
print(face_stats)
window_dominant_emotions = set() window_dominant_emotions = set()
# Determine dominant emotion for each face in the window # Determine dominant emotion for each face in the window
for _, counter in face_stats.items(): for _, counter in face_stats.items():
total_detections = sum(counter.values()) total_detections = sum(counter.values())
if total_detections >= min_frames_required: if total_detections >= self.min_frames_required:
dominant_emotion = counter.most_common(1)[0][0] dominant_emotion = counter.most_common(1)[0][0]
window_dominant_emotions.add(dominant_emotion) window_dominant_emotions.add(dominant_emotion)
await self.update_emotions(prev_dominant_emotions, window_dominant_emotions) await self.update_emotions(prev_dominant_emotions, window_dominant_emotions)
prev_dominant_emotions = window_dominant_emotions prev_dominant_emotions = window_dominant_emotions
face_stats.clear() face_stats.clear()
next_window_time = time.time() + window_duration next_window_time = time.time() + self.window_duration
except zmq.Again: except zmq.Again:
self.logger.warning("No video frame received within timeout.") self.logger.warning("No video frame received within timeout.")
@@ -112,16 +132,15 @@ class VisualEmotionRecognitionAgent(BaseAgent):
return return
emotion_beliefs_remove = [] emotion_beliefs_remove = []
# Remove emotions that have disappeared
for emotion in emotions_to_remove: for emotion in emotions_to_remove:
self.logger.info(f"Emotion '{emotion}' has disappeared.") self.logger.info(f"Emotion '{emotion}' has disappeared.")
try: try:
emotion_beliefs_remove.append(Belief(name="emotion_detected", arguments=[emotion], remove=True)) emotion_beliefs_remove.append(Belief(name="emotion_detected", arguments=[emotion],
remove=True))
except ValidationError: except ValidationError:
self.logger.warning("Invalid belief for emotion removal: %s", emotion) self.logger.warning("Invalid belief for emotion removal: %s", emotion)
emotion_beliefs_add = [] emotion_beliefs_add = []
# Add new emotions that have appeared
for emotion in emotions_to_add: for emotion in emotions_to_add:
self.logger.info(f"New emotion detected: '{emotion}'") self.logger.info(f"New emotion detected: '{emotion}'")
try: try:
@@ -131,7 +150,7 @@ class VisualEmotionRecognitionAgent(BaseAgent):
beliefs_list_add = [b.model_dump() for b in emotion_beliefs_add] beliefs_list_add = [b.model_dump() for b in emotion_beliefs_add]
beliefs_list_remove = [b.model_dump() for b in emotion_beliefs_remove] beliefs_list_remove = [b.model_dump() for b in emotion_beliefs_remove]
payload = {"create": beliefs_list_add, "delete": beliefs_list_remove, "replace": []} payload = {"create": beliefs_list_add, "delete": beliefs_list_remove}
message = InternalMessage( message = InternalMessage(
to=settings.agent_settings.bdi_core_name, to=settings.agent_settings.bdi_core_name,

View File

@@ -11,20 +11,28 @@ class VisualEmotionRecognizer(abc.ABC):
pass pass
@abc.abstractmethod @abc.abstractmethod
def sorted_dominant_emotions(self, image): def sorted_dominant_emotions(self, image) -> list[str]:
"""Recognize emotion from the given image. """
Recognize dominant emotions from faces in the given image.
Emotions can be one of ['angry', 'disgust', 'fear', 'happy', 'sad', 'surprise', 'neutral'].
To minimize false positives, consider filtering faces with low confidence.
:param image: The input image for emotion recognition. :param image: The input image for emotion recognition.
:return: Detected emotion label. :return: List of dominant emotion detected for each face in the image,
sorted per face.
""" """
pass pass
class DeepFaceEmotionRecognizer(VisualEmotionRecognizer): class DeepFaceEmotionRecognizer(VisualEmotionRecognizer):
"""
DeepFace-based implementation of VisualEmotionRecognizer.
DeepFape has proven to be quite a pessimistic model, so expect sad, fear and neutral
emotions to be over-represented.
"""
def __init__(self): def __init__(self):
self.load_model() self.load_model()
def load_model(self): def load_model(self):
# Initialize DeepFace model for emotion recognition
print("Loading Deepface Emotion Model...") print("Loading Deepface Emotion Model...")
dummy_img = np.zeros((224, 224, 3), dtype=np.uint8) dummy_img = np.zeros((224, 224, 3), dtype=np.uint8)
# analyze does not take a model as an argument, calling it once on a dummy image to load # analyze does not take a model as an argument, calling it once on a dummy image to load
@@ -32,7 +40,7 @@ class DeepFaceEmotionRecognizer(VisualEmotionRecognizer):
DeepFace.analyze(dummy_img, actions=['emotion'], enforce_detection=False) DeepFace.analyze(dummy_img, actions=['emotion'], enforce_detection=False)
print("Deepface Emotion Model loaded.") print("Deepface Emotion Model loaded.")
def sorted_dominant_emotions(self, image): def sorted_dominant_emotions(self, image) -> list[str]:
analysis = DeepFace.analyze(image, analysis = DeepFace.analyze(image,
actions=['emotion'], actions=['emotion'],
enforce_detection=False enforce_detection=False
@@ -41,12 +49,7 @@ class DeepFaceEmotionRecognizer(VisualEmotionRecognizer):
# Sort faces by x coordinate to maintain left-to-right order # Sort faces by x coordinate to maintain left-to-right order
analysis.sort(key=lambda face: face['region']['x']) analysis.sort(key=lambda face: face['region']['x'])
# Fear op 0, boost 0.2 aan happy, sad -0.1, neutral +0.1
analysis = [face for face in analysis if face['face_confidence'] >= 0.90] analysis = [face for face in analysis if face['face_confidence'] >= 0.90]
dominant_emotions = [face['dominant_emotion'] for face in analysis] dominant_emotions = [face['dominant_emotion'] for face in analysis]
return dominant_emotions return dominant_emotions

View File

@@ -78,6 +78,10 @@ class BehaviourSettings(BaseModel):
:ivar transcription_words_per_token: Estimated words per token for transcription timing. :ivar transcription_words_per_token: Estimated words per token for transcription timing.
:ivar transcription_token_buffer: Buffer for transcription tokens. :ivar transcription_token_buffer: Buffer for transcription tokens.
:ivar conversation_history_length_limit: The maximum amount of messages to extract beliefs from. :ivar conversation_history_length_limit: The maximum amount of messages to extract beliefs from.
:ivar visual_emotion_recognition_window_duration_s: Duration in seconds over which to aggregate
emotions and update emotion beliefs.
:ivar visual_emotion_recognition_min_frames_per_face: Minimum number of frames per face required
to consider a face valid.
""" """
# ATTENTION: When adding/removing settings, make sure to update the .env.example file # ATTENTION: When adding/removing settings, make sure to update the .env.example file
@@ -101,6 +105,9 @@ class BehaviourSettings(BaseModel):
# Text belief extractor settings # Text belief extractor settings
conversation_history_length_limit: int = 10 conversation_history_length_limit: int = 10
# Visual Emotion Recognition settings
visual_emotion_recognition_window_duration_s: int = 5
visual_emotion_recognition_min_frames_per_face: int = 3
class LLMSettings(BaseModel): class LLMSettings(BaseModel):
""" """