A Training-Free Approach for Generic Object Detection

Abstract

We present an approach for generic object detection using single query image for finding and locating visually similar objects from target images. The key challenge here is describing an object class using only one query without any training. Our approach is based on computation of Local Self-Similarity descriptors which captures local internal geometric layout within an image and is good representative of object class. We propose to use only predefined landmark points from query image which significantly improves performance of detection. We also present few novel ideas for selection of informative descriptors from the set of all descriptors of the test image to reduce computational expense in feature matching. The algorithm yields Hough-style similarity surface indicating likelihood of presence of the query object at every location. Presence and location of objects are finalized by employing two significance tests followed by non-maxima suppression. We evaluate results of the proposed approach on UIUC car dataset and achieve higher accuracy in comparison with earlier training-free approaches. Results are also analysed on ETHZ shape classes dataset which accounts for large intra-class variations, scale variations, and clutter. Performance of detection on this challenging dataset having diverse contexts demonstrates robustness and efficacy of the algorithm.

Keywords:

• Cosine similarity
• Generic
• Non-maxima suppression
• Object detection
• Self similarity
• Training

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This publication is an outcome of the R &D work undertaken project under the Visvesvaraya PhD Scheme of Ministry of Electronics & Information Technology, Government of India PhD-MLA/4(67)/2015-16, being implemented by Digital India Corporation.

Notes on contributors

Bhakti V. Baheti

Bhakti V Baheti is a PhD research scholar at Center of Excellence in Signal and Image Processing, SGGS Institute of Engineering and Technology, Nanded, Maharashtra, India. She received her BE in electronics and telecommunication engineering from Pune University, India in 2013 and MTech from SGGS Institute of Engineering and Technology, Nanded, India in 2015. She is a recipient of prestigious Visvesvaraya fellowship of Government of India towards PhD. Her major fields of interests include computer vision,machine learning, convolutional neural networks and advanced driver assistance systems. Corresponding author. Email: [email protected]

Sanjay N. Talbar

Sanjay N Talbar is a professor in Electronics and Telecommunication Engineering department at SGGS Institute of Engineering and Technology, Nanded, Maharashtra, India. He received BE and ME in electronics engineering from the Marathwada University, Aurangabad, India, in 1985 and in 1990, respectively, and PhD from S R T Marathwada University, Nanded in May 2000. He worked at various levels in academic, research and professional organizations. He had collaborative research programme at Cardiff University Wales, UK. He is an author of 10 books including 3 books by Springer. He has published several research papers in reputed journals and about 130 papers in conferences. His areas of interest include image processing, multimedia computing systems, real time embedded system and design, pattern recognition, machine learning and deep learning. He is a member of several professional societies like IEEE, AMPI, IET, a fellow of IETE and a life member of ISTE. Email: [email protected]

Suhas S. Gajre

Suhas S Gajre is working as an associate professor in the department of Electronics and Telecommunications Engineering at SGGS Institute of Engineering and Technology Nanded, Maharashtra, India. He received BE (Electronics) and ME (Electronics) from Shri Guru Gobind Singhji Institute of Engineering and Technology, Vishnupuri, Nanded, Maharashtra State, India in 1990 and 1995, respectively. He obtained his PhD from Indian Institute of Technology, Delhi, India, in biomedical engineering in 2007. His research interests include Biomedical signal and image processing, pattern recognition and analog and mixed signal VLSI design. He is a Fellow of IETE, a Fellow of IE(I), a member of IET and a life member of ISTE. Email: [email protected]