Srikrishna Karanam

I am a Research Scientist in the Vision group at Siemens Corporate Technology, where I work on Computer Vision and related problems. I earned my Ph.D. in Computer and Systems Engineering at Rensselaer Polytechnic Institute, where my adviser was Prof. Rich Radke.

For an overview of my Ph.D. work, please see this article from ALERT@Northeastern.

Email  /  CV

News
Research

I am interested in computer vision, video processing, machine learning and optimization. I am particularly interested in video analytics problems as they occur in large networks of cameras.

karanam-gou-arxiv16

A Systematic Evaluation and Benchmark for Person Re-Identification: Features, Metrics, and Datasets
Srikrishna Karanam*, Mengran Gou*, Ziyan Wu, Angels Rates-Borras, Octavia Camps, Richard J. Radke
IEEE Transactions on Pattern Analysis and Machine Intelligence (T-PAMI), accepted, Feb 2018.
* equal contribution
code / supplemental material

We are conducting a systematic study of existing features, metric learning, and multi-shot ranking algorithms for re-id. Please also see this collection and review of re-id datasets by Mengran.

conceptGAN

Learning Compositional Visual Concepts with Mutual Consistency
Yunye Gong, Srikrishna Karanam, Ziyan Wu, Kuan-Chuan Peng, Jan Ernst, Peter C. Doerschuk
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2018   (Spotlight)

We propose ConceptGAN, a framework that can jointly learn, transfer and compose concepts to generate semantically meaningful images, even in subdomains with no training data.

conceptGAN

End-to-end learning of keypoint detector and descriptor for pose invariant 3D matching
Georgios Georgakis, Srikrishna Karanam, Ziyan Wu, Jan Ernst, Jana Kosecka
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2018

We propose an end-to-end deep network for jointly learning keypoint detection and description from only synthetic data in 3D. The model is a Siamese architecture that integrates Faster R-CNN to generate proposals for validation by a contrastive loss.

karanam-arxiv17

Rank Persistence: Assessing the Temporal Performance of Real-World Person Re-Identification
Srikrishna Karanam, Eric Lam, Richard J. Radke
International Conference on Distributed Smart Cameras (ICDSC), Sept. 2017

We present a new evaluation methodology that explicitly considers practical aspects involved when deploying a re-id algorithm in the real world.

karanam-csvt17

Learning Affine Hull Representations for Multi-Shot Person Re-Identification
Srikrishna Karanam, Ziyan Wu, Richard J. Radke
IEEE Transactions on Circuits and Systems for Video Technology (T-CSVT), accepted July 2017.

We tackle the multi-shot re-id problem by learning discriminative representations using affine hulls of data and show improvements with existing metric learning algorithms.

gou-cvprw17

DukeMTMC4ReID: A Large-Scale Multi-Camera Person Re-Identification Dataset
Mengran Gou, Srikrishna Karanam, Wenqian Liu, Octavia Camps, Richard J. Radke
Workshop on Target Re-Identification and Multi-Target Multi-Camera Tracking, CVPR 2017
data

We introduce a new, large-scale, dataset for re-id based on the DukeMTMC multi-target tracking dataset.

karanam-ivc16

Person Re-Identification with Block Sparse Recovery
Srikrishna Karanam, Yang Li, Richard J. Radke
Elsevier Image and Vision Computing (IVC), accepted, Feb. 2017

We formulate multi-shot re-identification as a block sparse recovery problem. This subsumes our CVPR-W 2015 paper.

karanam-csvt16

From the Lab to the Real World: Re-Identification in an Airport Camera Network
Octavia Camps, Mengran Gou, Tom Hebble, Srikrishna Karanam, Oliver Lehmann, Yang Li, Richard J. Radke, Ziyan Wu, Fei Xiong
IEEE Transactions on Circuits and Systems for Video Technology (T-CSVT), accepted, Dec. 2016

This paper describes our real-time end-to-end person re-identification system. This subsumes our ICDSC 2014 paper.

karanam-iccv15

Person Re-Identification with Discriminatively Trained Viewpoint Invariant Dictionaries
Srikrishna Karanam, Yang Li, Richard J. Radke
IEEE International Conference on Computer Vision (ICCV), 2015
spotlight / code / CMC

We learn a dictionary capable of discriminatively and sparsely encoding gallery and probe features.

karanam-bmvc15

Particle Dynamics and Multi-Channel Feature Dictionaries for Robust Visual Tracking
Srikrishna Karanam, Yang Li, Richard J. Radke
British Machine Vision Conference (BMVC), 2015   (Oral Presentation)
supplement / videos / slides / code

We construct multi-channel feature dictionaries as part of the target appearance model and exploit particle dynamical information to improve tracking accuracy.

li-bmvc15

Multi-Shot Human Re-Identification Using Adaptive Fisher Discriminant Analysis
Yang Li, Ziyan Wu, Srikrishna Karanam, Richard J. Radke
British Machine Vision Conference (BMVC), 2015

We combine Fisher discriminant analysis and hierarchical image sequence clustering to adaptively learn a discriminative feature space.

karanam-msf15

Sparse Re-Id: Block Sparsity for Person Re-Identification
Srikrishna Karanam, Yang Li, Richard J. Radke
Workshop on Multi-Sensor Fusion for Dynamic Scene Understanding, CVPR 2015

We formulate multi-shot re-identification as a block sparse recovery problem.

li-icsdc15

Real-World Re-Identification in an Airport Camera Network
Yang Li, Ziyan Wu, Srikrishna Karanam, Richard J. Radke
International Conference on Distributed Smart Cameras (ICDSC), Nov. 2014
Demo video

This paper describes a preliminary version of our real-time end-to-end person re-identification system.


Template credits