Object Detection Demistified-State of art DeepNets
saiamrit |
Description:
90% of data in the internet today is either image or video.The exponential rise of visual data has continuously urged researchers to develop robust and efficient Object detection algorithms,but CNN or R-CNN or YOLO or SSD which algorithm can give best results.In this talk I will try to cover salient features in some of the most influential works in this problem statement.The talk begins with intro to CNNs and goes into detailed discussion of state-of-the-art deep learning algorithms used for object detection.
Structure of the talk - The talk is structured into 3 sections : In the first 20 minutes we will have a talk on the architectures, then 10 minutes will be dedicated for some hands-on demo to build a CNN using Keras/Pytorch and the rest of the time will be for QnA.
Contents - The talk will begin with a discussion on Convolution Neural Networks and various terms associated like Convolution,pooling,activation used etc and there after discussing about the various state-of-the-art algorithms like R-CNN,Fast R-CNN,Faster R-CNN,R-FCN,YOLO and SSD.One of my analysis criteria will be on their speed at inference allowing real-time analysis.
Take aways :
- What is a CNN,what are convolution,pooling etc.
- What are R-CNN,Fast R-CNN,Faster R-CNN,R-FCN,YOLO and SSD
- How to implement a CNN using keras/Pytorch.
Prerequisites:
- Basic python or any other language programming.
- Basic knowledge of Machine Learning and Neural Networks.
- Most importantly an interest to learn a new concept.
Content URLs:
To be updated soon !!
Speaker Info:
The speaker is a 4th year undergraduate student from the department of Computer Science and Engineering at IIIT Bhubaneswar. He is a Data science, Machine Learning and Deep learning enthusiast.He has an experience of over 2 years in this field and has worked on Machine Learning and Deep Learning and it's application to Computer Vision(CV) and Natural Language Processing(NLP). He has worked on few self projects and been a part of 2 research Internships, One at IIIT Bangalore and another at IIT Kharagpur . He has experience of working with various libraries like sci-kit ,Tensorflow ,Keras ,Torch and Pytorch.