Object Detection Tutorial#
 Run on Google Colab
|
 View source on GitHub
|
 Download notebook
|
Connect to EvaDB#
%pip install --quiet "evadb[vision,notebook]"
import evadb
cursor = evadb.connect().cursor()
Note: you may need to restart the kernel to use updated packages.
Download the Surveillance Videos#
In this tutorial, we focus on the UA-DETRAC benchmark. UA-DETRAC is a challenging real-world multi-object detection and multi-object tracking benchmark. The dataset consists of 10 hours of videos captured with a Cannon EOS 550D camera at 24 different locations at Beijing and Tianjin in China.
# Getting the video files
!wget -nc "https://www.dropbox.com/s/k00wge9exwkfxz6/ua_detrac.mp4?raw=1" -O ua_detrac.mp4
File ‘ua_detrac.mp4’ already there; not retrieving.
Load the videos for analysis#
We use a regular expression to load all the videos into the table in a single command
cursor.query("DROP TABLE IF EXISTS ObjectDetectionVideos;").df()
cursor.load(file_regex='ua_detrac.mp4', format="VIDEO", table_name='ObjectDetectionVideos').df()
| 0 | |
|---|---|
| 0 | Number of loaded VIDEO: 1 |
Register YOLO Object Detector as an User-Defined Function (UDF) in EvaDB#
cursor.query("""
CREATE UDF IF NOT EXISTS Yolo
TYPE ultralytics
'model' 'yolov8m.pt';
""").df()
| 0 | |
|---|---|
| 0 | UDF Yolo already exists, nothing added. |
Run the YOLO Object Detector on the video#
yolo_query = cursor.table("ObjectDetectionVideos")
yolo_query = yolo_query.filter("id < 20")
yolo_query = yolo_query.select("id, Yolo(data)")
response = yolo_query.df()
print(response)
objectdetectionvideos.id \
0 0
1 1
2 2
3 3
4 4
5 5
6 6
7 7
8 8
9 9
10 10
11 11
12 12
13 13
14 14
15 15
16 16
17 17
18 18
19 19
yolo.labels \
0 [car, car, car, car, car, car, person, car, ca...
1 [car, car, car, car, car, car, car, car, perso...
2 [car, car, car, person, car, car, car, car, ca...
3 [car, car, car, car, car, car, person, car, ca...
4 [car, car, car, car, car, car, person, car, ca...
5 [car, car, car, car, car, car, car, person, ca...
6 [car, car, car, car, person, car, car, car, ca...
7 [car, car, car, car, car, car, car, car, perso...
8 [car, car, car, car, car, car, car, car, perso...
9 [car, car, car, person, car, car, car, car, ca...
10 [car, car, person, car, car, car, car, car, ca...
11 [car, car, car, person, car, car, truck, car, ...
12 [car, car, person, car, car, car, car, car, ca...
13 [car, car, car, person, car, truck, car, car, ...
14 [car, car, car, car, person, car, car, car, ca...
15 [car, car, car, car, person, car, car, car, ca...
16 [car, car, car, person, car, car, car, car, ca...
17 [car, car, car, car, person, car, car, car, ca...
18 [car, car, person, car, car, car, car, car, ca...
19 [car, person, car, car, car, car, car, car, ca...
yolo.bboxes \
0 [[827.97412109375, 274.3556213378906, 959.1490...
1 [[155.66152954101562, 477.1302795410156, 275.7...
2 [[835.0940551757812, 276.74273681640625, 959.4...
3 [[839.3997802734375, 278.81060791015625, 959.3...
4 [[843.2724609375, 280.5651550292969, 959.40820...
5 [[847.2899780273438, 281.1034240722656, 959.29...
6 [[851.1822509765625, 282.2825927734375, 959.58...
7 [[854.6375732421875, 282.7655029296875, 959.54...
8 [[859.2888793945312, 283.25140380859375, 959.5...
9 [[863.5880126953125, 285.3915710449219, 959.52...
10 [[867.6785888671875, 286.24920654296875, 959.5...
11 [[871.6613159179688, 286.8082275390625, 959.45...
12 [[634.4969482421875, 223.8955535888672, 751.20...
13 [[880.8702392578125, 288.19842529296875, 959.5...
14 [[885.093017578125, 288.404052734375, 959.4323...
15 [[889.7506713867188, 290.65985107421875, 959.4...
16 [[894.7098999023438, 292.442138671875, 959.437...
17 [[899.2138671875, 293.09765625, 959.4815673828...
18 [[184.9110565185547, 388.7672119140625, 300.05...
19 [[649.16552734375, 228.12881469726562, 770.541...
yolo.scores
0 [0.76, 0.75, 0.74, 0.7, 0.67, 0.66, 0.61, 0.61...
1 [0.75, 0.74, 0.7, 0.68, 0.68, 0.66, 0.66, 0.59...
2 [0.84, 0.78, 0.71, 0.68, 0.67, 0.67, 0.66, 0.6...
3 [0.83, 0.77, 0.68, 0.67, 0.66, 0.66, 0.64, 0.6...
4 [0.8, 0.74, 0.73, 0.7, 0.68, 0.67, 0.65, 0.63,...
5 [0.83, 0.76, 0.75, 0.74, 0.67, 0.64, 0.62, 0.6...
6 [0.83, 0.77, 0.74, 0.71, 0.63, 0.63, 0.62, 0.6...
7 [0.82, 0.77, 0.67, 0.65, 0.65, 0.63, 0.62, 0.5...
8 [0.79, 0.78, 0.65, 0.63, 0.62, 0.62, 0.6, 0.59...
9 [0.83, 0.75, 0.66, 0.66, 0.65, 0.63, 0.61, 0.5...
10 [0.76, 0.74, 0.7, 0.67, 0.62, 0.6, 0.54, 0.53,...
11 [0.81, 0.74, 0.68, 0.66, 0.61, 0.57, 0.55, 0.5...
12 [0.8, 0.77, 0.61, 0.59, 0.58, 0.56, 0.55, 0.54...
13 [0.8, 0.76, 0.72, 0.69, 0.6, 0.6, 0.55, 0.55, ...
14 [0.83, 0.76, 0.75, 0.64, 0.64, 0.61, 0.61, 0.5...
15 [0.84, 0.77, 0.76, 0.69, 0.68, 0.63, 0.62, 0.6...
16 [0.81, 0.73, 0.69, 0.68, 0.68, 0.65, 0.6, 0.6,...
17 [0.76, 0.75, 0.68, 0.67, 0.66, 0.64, 0.61, 0.5...
18 [0.79, 0.76, 0.72, 0.67, 0.64, 0.58, 0.57, 0.5...
19 [0.78, 0.74, 0.74, 0.64, 0.61, 0.56, 0.55, 0.5...
Visualizing output of the Object Detector on the video#
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[['yolo.bboxes', 'yolo.labels']][df.index == frame_id]
if df.size:
dfLst = df.values.tolist()
for bbox, label in zip(dfLst[0][0], dfLst[0][1]):
x1, y1, x2, y2 = bbox
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)
# frame label
cv2.putText(frame, 'Frame ID: ' + str(frame_id), (700, 500), cv2.FONT_HERSHEY_SIMPLEX, 1.2, color2, thickness)
video.write(frame)
# Stop after twenty frames (id < 20 in previous query)
if frame_id == 20:
break
# 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 = 'ua_detrac.mp4'
output_path = 'video.mp4'
annotate_video(response, input_path, output_path)
Video.from_file(output_path)
Drop the function if needed#
cursor.query("DROP UDF IF EXISTS Yolo").df()
| 0 | |
|---|---|
| 0 | UDF Yolo successfully dropped |