Department of Electrical and Electronics Engineering

Permanent URI for this collectionhttp://localhost:4000/handle/123456789/1925

Browse

Has files: NoSubject: search.filters.subject.CNN

Search Results

Now showing 1 - 3 of 3
  • No Thumbnail Available
    Item
    An FPGA Based Hardware Accelerator for Classification of Handwritten Digits
    (Springer, 2019-04) Chaturvedi, Nitin
    Over the past few years, Convolutional Neural Networks (CNNs) have provided major breakthroughs in fields such as computer vision and natural language processing, resulting in a rise in the adoption of CNNs with increased levels of complexity. Consequently, the need for fast and power efficient processing of such networks has become critically important. Conventional hardware solutions, namely CPUs and GPUs, fail to address these requirements as CPUs are not suited for processing massively parallel multiply and accumulate (MAC) operations and GPUs are not power efficient. However, Field Programmable Gate Arrays (FPGAs) have emerged as a promising alternative due to their extensive programmability, ease of executing parallel operations and wide interfacing capabilities. In this paper, we have designed a hardware accelerator for speeding up the inference phase of LeNet–5 to enable faster classification of handwritten digits. We employ software quantization for ease of implementation on FPGAs, and partial pipelining to process the various layers of a typical CNN. Targeting the Xilinx Zynq-7000 SoC, we report a speedup improvement of at least 4.7x and power efficiency improvement of 32x compared to similar works.
  • No Thumbnail Available
    Item
    Comparative Performance Study of CNN-based Algorithms and YOLO
    (IEEE, 2022) Bitragunta, Sainath
    Tasks such as image classification, object detection, to mention a few, play an important role in computer vision. Numerous algorithms have been developed to improve the performance of such tasks for benchmark datasets. Although advanced algorithms offer state-of-the-art performance on such tasks, it is also important to analyze their algorithmic feasibility over the time to make it practical for end-user applications. This paper analyzes two such groups of algorithms, namely, Convolutional Neural Networks (CNN) based algorithms with You Only Look Once (YOLO) in terms of speed and accuracy.
  • No Thumbnail Available
    Item
    Scanned to Digital Face Images Matching With Siamese Network
    (IEEE, 2018) Gupta, Karunesh Kumar; Tiwari, Kamlesh
    Often in law enforcement and forensic application it is needed to match scanned facial image with a digital face image. This is because in many scenario, non-digital face images are obtained from the crime scene, news articles etc. that are needed to be identified. Non-digital face images are first scanned and then enhanced to match against the database. Challenges arrives because of poor quality of non-digital image, artifacts introduced in scanning process and high saturation etc. therefore matching becomes difficult. The methods used in literature involve specialized hand crafted pre-processing. In our paper, we propose an automated way of matching by using Siamese networks. The proposed method have been able to achieve an EER of 2.346% that is better than the current state-of-the-art.