JINKA
UNIVERSITY
COLLEGE OF NATURAL
AND COMPUTITIONAL SCIENCE
DEPARTMENT OF COMPUTER SCIENCE
REAL-TIME OBJECT DETECTION USING DEEP LEARNING
GROUP 7
• OUT LINE
• INTRODUCTION TO OBJECT DETECTION IN DEEP LEARNING
• METHODOLOGIES
• RESULTS
INTRODUCTION TO OBJECT DETECTION
• OBJECT DETECTION: IS A CRUCIAL ASPECT OF COMPUTER VISION, WITH APPLICATIONS IN VARIOUS
DOMAINS SUCH AS SURVEILLANCE, AUTONOMOUS DRIVING, AND MEDICAL IMAGING.
• IS A CRITICAL TASK IN VARIOUS APPLICATIONS SUCH AS AUTONOMOUS DRIVING, SURVEILLANCE, AND
ROBOTICS.
• THE ADVENT OF DEEP LEARNING HAS SIGNIFICANTLY ADVANCED THE CAPABILITIES OF OBJECT
DETECTION MODELS, ALLOWING FOR REAL-TIME PROCESSING AND HIGHER ACCURACY.
• THE ROLE OF DEEP LEARING IN THIS CASE IS TO IMPROVING FEATURE EXTRACTION, ENABLING
EFFICIENT REGION PROPOSAL AND ACCURATE CLASSIFICATION AND LOCALIZATION , SIMPLIFIYING
THE TRAINING PROCESS AND HANDLING COMPLEX SCENARIOS EFFECTIVELY.
• OBJECTIVE(S) AND SCOPE
OBJECTIVES
TO DEVELOP A DEEP LEARNING MODEL FOR REAL-TIME OBJECT DETECTION.
TO ACHIEVE HIGH ACCURACY AND LOW LATENCY IN OBJECT DETECTION.
TO EVALUATE THE MODEL USING STANDARD METRICS AND DATASETS.
SCOPE
THIS REPORT FOCUSES ON THE DEVELOPMENT, TRAINING, AND EVALUATION OF A DEEP
LEARNING-BASED OBJECT DETECTION MODEL. THE SCOPE INCLUDES DATA
COLLECTION, PREPROCESSING, MODEL SELECTION, TRAINING, TESTING, AND
PERFORMANCE ANALYSIS.
LIMITATIONS
• THIS PROJECT WILL NOT WORK ON VIDEO STREAM
• ONLY DETECT ON CLEAR IMAGE
• NOT FAST ENOUGH AS OTHER
• HAVE LOW PERFORMANCE
METHODOLOGY
THE METHODOLOGY INCLUDES: -
DATA COLLECTION AND PREPROCESSING.
MODEL SELECTION AND TRAINING.
PERFORMANCE EVALUATION.
DATA COLLECTION :COLLECTING DATA FOR TRAINING AND TEST METHODS THERE ARE MANY
RESOURCES FOR THIS PURPOSE .
SUCH AS
KALGLE.COM
COCO.NAMES
DATA PREPROCESSING : PREPARE THE COLLECTED DATA FOR THE MODEL MEAN THAT
ANALYZING THE DATA ,CLEAR THE UNWANTED DATA, CONVERT TO STANDARD FORMAT AND
ORGANIZE THE DATA .
• CONT….
• MODEL SELECTION : CHOOSING THE FUNCTIONAL MODEL FOR OBJECT DETECTION THERE
ARE MANY PRE TRAINED MODEL LIKE:-
YOLO(YOU ONLY LOOK ONCE )
SSD(SINGLE SHOT MULTIBOX)
FROZEN_MODEL
• MODEL TRAINING: THIS STAGE IS FEEDING THE MODEL BY THE COLLECTED AND
ORGANIZED DATA AND TEST THE MODEL IF IS WAS DETECTING CORRECTLY .
• AT THIS STAGE THE MODEL TRAIN USING OR BASE ON THE LABELED DATA LIST .
• PERFORMANCE EVALUATION: CHECKING THE ACCURACY OF HOW MUCH IT DETECT
DIFFERENT OBJECT EFFECTIVELY FOR THIS STAGE THERE ARE MANY EVALUATION METRICS
LIKE :-
ACCURACY
F1_SCORE ….. AND SO FOR
RESULT
• STRENGTHS OF THE MODEL
THIS MODEL HAVE SOME STRENGTH IT DETCET OBJECT WELL WHEN WE HAVE CLEAR AND
WELL SEEN IMAGES ALSO IT HAVE GOOD ENOUGH PREFORMACE ON DETECTING OBJECT
• WEAKNESS OF THE MODEL
NO DETECT FORM VIDEO
NO DETECT FORM UNSEED OR UNCLEAR IMAGE
HAVE LOW PERFORMANCE IN CROWDED IMAGE EVEN IF THE IMAGE WAS CLEAR
O N
T I
E N
T T
R A
O U
R Y
F O
O U
Y
NK
H A
T
