EMOTION ANALYSIS#
Run on Google Colab | View source on GitHub | Download notebook |
Connect to EvaDB#
%pip install --quiet "evadb[vision,notebook]"
%pip install --quiet facenet_pytorch
import evadb
cursor = evadb.connect().cursor()
Note: you may need to restart the kernel to use updated packages.
Note: you may need to restart the kernel to use updated packages.
Download Video#
# A video of a happy person
!wget -nc "https://www.dropbox.com/s/gzfhwmib7u804zy/defhappy.mp4?raw=1" -O defhappy.mp4
# Adding Emotion detection
!wget -nc https://raw.githubusercontent.com/georgia-tech-db/eva/master/evadb/udfs/emotion_detector.py
# Adding Face Detector
!wget -nc https://raw.githubusercontent.com/georgia-tech-db/eva/master/evadb/udfs/face_detector.py
--2023-06-17 00:21:00-- https://www.dropbox.com/s/gzfhwmib7u804zy/defhappy.mp4?raw=1
Resolving www.dropbox.com (www.dropbox.com)... 162.125.9.18, 2620:100:601f:18::a27d:912
Connecting to www.dropbox.com (www.dropbox.com)|162.125.9.18|:443... connected.
HTTP request sent, awaiting response... 302 Found
Location: /s/raw/gzfhwmib7u804zy/defhappy.mp4 [following]
--2023-06-17 00:21:00-- https://www.dropbox.com/s/raw/gzfhwmib7u804zy/defhappy.mp4
Reusing existing connection to www.dropbox.com:443.
HTTP request sent, awaiting response... 302 Found
Location: https://ucdb88e4f7cbe6b7fcce2afc862e.dl.dropboxusercontent.com/cd/0/inline/B-IdFWkGulxDd5XDfmkDmrgSyuBLmJALJZOejL8n7U0pq8hL006_iYe-Xoz6IzwXZCFF0cGSeeldLr0zjsMAKVWNeXc-v3c9V_t-GYMzeb7RNu6gGi3QH9eEFCii811R_z3JehT0M1_YMiqVMHIekrYju7uZFcccji1SUoekefqnRA/file# [following]
--2023-06-17 00:21:00-- https://ucdb88e4f7cbe6b7fcce2afc862e.dl.dropboxusercontent.com/cd/0/inline/B-IdFWkGulxDd5XDfmkDmrgSyuBLmJALJZOejL8n7U0pq8hL006_iYe-Xoz6IzwXZCFF0cGSeeldLr0zjsMAKVWNeXc-v3c9V_t-GYMzeb7RNu6gGi3QH9eEFCii811R_z3JehT0M1_YMiqVMHIekrYju7uZFcccji1SUoekefqnRA/file
Resolving ucdb88e4f7cbe6b7fcce2afc862e.dl.dropboxusercontent.com (ucdb88e4f7cbe6b7fcce2afc862e.dl.dropboxusercontent.com)... 162.125.9.15, 2620:100:601f:15::a27d:90f
Connecting to ucdb88e4f7cbe6b7fcce2afc862e.dl.dropboxusercontent.com (ucdb88e4f7cbe6b7fcce2afc862e.dl.dropboxusercontent.com)|162.125.9.15|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 2699034 (2.6M) [video/mp4]
Saving to: 'defhappy.mp4'
defhappy.mp4 100%[===================>] 2.57M --.-KB/s in 0.02s
2023-06-17 00:21:01 (145 MB/s) - 'defhappy.mp4' saved [2699034/2699034]
File 'emotion_detector.py' already there; not retrieving.
File 'face_detector.py' already there; not retrieving.
Load video for analysis#
response = cursor.query("DROP TABLE IF EXISTS HAPPY;").df()
print(response)
cursor.load(file_regex="defhappy.mp4", table_name="HAPPY", format="VIDEO").df()
06-17-2023 00:21:02 WARNING[drop_object_executor:drop_object_executor.py:_handle_drop_table:0050] Table: HAPPY does not exist
0
0 Table: HAPPY does not exist
0 | |
---|---|
0 | Number of loaded VIDEO: 1 |
Create a function for analyzing the frames#
cursor.query("""CREATE UDF IF NOT EXISTS EmotionDetector
INPUT (frame NDARRAY UINT8(3, ANYDIM, ANYDIM))
OUTPUT (labels NDARRAY STR(ANYDIM), scores NDARRAY FLOAT32(ANYDIM))
TYPE Classification IMPL 'emotion_detector.py';
""").df()
cursor.query("""CREATE UDF IF NOT EXISTS FaceDetector
INPUT (frame NDARRAY UINT8(3, ANYDIM, ANYDIM))
OUTPUT (bboxes NDARRAY FLOAT32(ANYDIM, 4),
scores NDARRAY FLOAT32(ANYDIM))
TYPE FaceDetection
IMPL 'face_detector.py';
""").df()
0 | |
---|---|
0 | UDF FaceDetector already exists, nothing added. |
Run the Face Detection UDF on video#
query = cursor.table("HAPPY")
query = query.filter("id < 10")
query = query.select("id, FaceDetector(data)")
query.df()
2023-06-17 00:21:05,712 INFO worker.py:1625 -- Started a local Ray instance.
happy.id | facedetector.bboxes | facedetector.scores | |
---|---|---|---|
0 | 0 | [[502, 94, 762, 435], [238, 296, 325, 398]] | [0.99990165, 0.79820246] |
1 | 1 | [[501, 96, 763, 435]] | [0.999918] |
2 | 2 | [[504, 97, 766, 437]] | [0.9999138] |
3 | 3 | [[498, 90, 776, 446]] | [0.99996686] |
4 | 4 | [[496, 99, 767, 444]] | [0.9999982] |
5 | 5 | [[499, 87, 777, 448], [236, 305, 324, 407]] | [0.9999136, 0.8369736] |
6 | 6 | [[500, 89, 778, 449]] | [0.9999131] |
7 | 7 | [[501, 89, 781, 452]] | [0.9999124] |
8 | 8 | [[503, 90, 783, 450]] | [0.99994683] |
9 | 9 | [[508, 87, 786, 447]] | [0.999949] |
Run the Emotion Detection UDF on the outputs of the Face Detection UDF#
query = cursor.table("HAPPY")
query = query.cross_apply("UNNEST(FaceDetector(data))", "Face(bbox, conf)")
query = query.filter("id < 15")
query = query.select("id, bbox, EmotionDetector(Crop(data, bbox))")
response = query.df()
import cv2
from pprint import pprint
from matplotlib import pyplot as plt
def annotate_video(detections, input_video_path, output_video_path):
color1=(207, 248, 64)
color2=(255, 49, 49)
thickness=4
vcap = cv2.VideoCapture(input_video_path)
width = int(vcap.get(3))
height = int(vcap.get(4))
fps = vcap.get(5)
fourcc = cv2.VideoWriter_fourcc('m', 'p', '4', 'v') #codec
video=cv2.VideoWriter(output_video_path, fourcc, fps, (width,height))
frame_id = 0
# Capture frame-by-frame
# ret = 1 if the video is captured; frame is the image
ret, frame = vcap.read()
while ret:
df = detections
df = df[['Face.bbox', 'emotiondetector.labels', 'emotiondetector.scores']][df.index == frame_id]
if df.size:
x1, y1, x2, y2 = df['Face.bbox'].values[0]
label = df['emotiondetector.labels'].values[0]
score = df['emotiondetector.scores'].values[0]
x1, y1, x2, y2 = int(x1), int(y1), int(x2), int(y2)
# object bbox
frame=cv2.rectangle(frame, (x1, y1), (x2, y2), color1, thickness)
# object label
cv2.putText(frame, label, (x1, y1-10), cv2.FONT_HERSHEY_SIMPLEX, 0.9, color1, thickness)
# object score
cv2.putText(frame, str(round(score, 5)), (x1+120, y1-10), cv2.FONT_HERSHEY_SIMPLEX, 0.9, color1, thickness)
# frame label
cv2.putText(frame, 'Frame ID: ' + str(frame_id), (700, 500), cv2.FONT_HERSHEY_SIMPLEX, 1.2, color2, thickness)
video.write(frame)
# Show every fifth frame
if frame_id % 5 == 0:
plt.imshow(frame)
plt.show()
frame_id+=1
ret, frame = vcap.read()
video.release()
vcap.release()
from ipywidgets import Video, Image
input_path = 'defhappy.mp4'
output_path = 'video.mp4'
annotate_video(response, input_path, output_path)
Dropping an User-Defined Function (UDF)#
#cursor.drop(item_name="EmotionDetector", item_type="UDF").df()
#cursor.drop(item_name="FaceDetector", item_type="UDF").df()