Realtime Object Detection And Localization In Compressive Sensed Video

• doi: 10.1109/ICIP42928.2021.9506769
• title: Real-Time Object Detection and Localization in Compressive Sensed Video
• publisher: IEEE
• isbn: 978-1-6654-3102-6
• issn: 1522-4880
• rank: 3397
• access_type: LOCKED
• content_type: Conferences
• abstract: Typically a 1-2MP CCTV camera generates around 7-12GB of data per day. Frame-by-frame processing of such an enormous amount of data requires hefty computational resources. In recent years, compressive sensing approaches have shown impressive compression results by reducing the sampling bandwidth. Different sampling mechanisms were developed to incorporate compressive sensing in image and video acquisition. Though all-CMOS [1], [2] sensor cameras that perform compressive sensing can help save a lot of bandwidth on sampling and minimize the memory required to store videos, the traditional signal processing, and deep learning models can realize operations only on the reconstructed data. To realize the original uncompressed domain, most reconstruction techniques are computationally expensive and time-consuming. To bridge this gap, we propose a novel task of detection and localization of objects directly on the compressed frames. Thereby mitigating the need to reconstruct the frames and reducing the search rate up to $20 \times$ (compression rate). We achieved an accuracy of 46.27% mAP with the proposed model on a GeForce GTX 1080 Ti. We were also able to show real-time inference on an NVIDIA TX2 embedded board with 45.11% mAP, thereby achieving the best balance between the accuracy, inference time, and memory constraints.
• article_number: 9506769
• pdf_url: https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9506769
• html_url: https://ieeexplore.ieee.org/document/9506769/
• abstract_url: https://ieeexplore.ieee.org/document/9506769/
• publication_title: 2021 IEEE International Conference on Image Processing (ICIP)
• conference_location: Anchorage, AK, USA
• conference_dates: 19-22 Sept. 2021
• publication_number: 9506008
• is_number: 9506009
• publication_year: 2021
• publication_date: 19-22 Sept. 2021
• start_page: 1489
• end_page: 1493
• citing_paper_count: 1
• citing_patent_count: 0
• download_count: 402
• insert_date: 20210823
• index_terms:
  • ieee_terms:
    • Location awareness
    • Image coding
    • Memory management
    • Object detection
    • Streaming media
    • Cameras
    • Real-time systems
  • author_terms:
    • Object Detection and Localization
    • Deep learning
    • Compressive Sensing
  • dynamic_index_terms:
    • Object Detection
    • Object Location
    • Real-time Object Detection
    • Real Time Object Detection
    • Detection Task
    • Localization Task
    • Postural Task
    • Compression Rate
    • Original Domain
    • Real-time Inference
    • Spatial Resolution
    • Temporal Dimension
    • Time Dimension
    • Temporal Properties
    • Intersection Over Union
    • Jaccard Similarity
    • Jaccard Distance
    • Intersection-over-union
    • Jaccard Index
    • Bounding Box
    • Average Precision
    • Mean Average Precision
    • Data Framework
    • Orthogonal Matching Pursuit
    • Video Compression
    • Sensor Pixel
• isbn_formats:
  • format: Print on Demand(PoD) ISBN, value: 978-1-6654-3102-6, isbnType: New-2005
  • format: Electronic ISBN, value: 978-1-6654-4115-5, isbnType: New-2005
• authors:
  • Author Name: Yeshwanth Ravi Theja Bethi
    Affiliation: Indian Institute of Science, Bangalore, India
    Author URL: https://ieeexplore.ieee.org/author/37086882310
    ID: 37086882310
    Order: 1
    Author Affiliations:
    • Indian Institute of Science, Bangalore, India
  • Author Name: Sathyaprakash Narayanan
    Affiliation: Indian Institute of Science, Bangalore, India
    Author URL: https://ieeexplore.ieee.org/author/37086886703
    ID: 37086886703
    Order: 2
    Author Affiliations:
    • Indian Institute of Science, Bangalore, India
  • Author Name: Venkat Rangan
    Affiliation: tinyVision.ai, San Diego, USA
    Author URL: https://ieeexplore.ieee.org/author/38179899400
    ID: 38179899400
    Order: 3
    Author Affiliations:
    • tinyVision.ai, San Diego, USA
  • Author Name: Anirban Chakraborty
    Affiliation: Indian Institute of Science, Bangalore, India
    Author URL: https://ieeexplore.ieee.org/author/37086881136
    ID: 37086881136
    Order: 4
    Author Affiliations:
    • Indian Institute of Science, Bangalore, India
  • Author Name: Chetan Singh Thakur
    Affiliation: Indian Institute of Science, Bangalore, India
    Author URL: https://ieeexplore.ieee.org/author/37085432865
    ID: 37085432865
    Order: 5
    Author Affiliations:
    • Indian Institute of Science, Bangalore, India

Image Sensor

• sensor_type: CMOS
• resolution: 1-2MP
• dynamic_range: Not specified
• pixel_size: Not specified
• dark_current: Not specified

Optical Data

• focal_length: Not specified
• aperture: Not specified
• field_of_view: Not specified
• distortion: Not specified

Performance Metrics

• frame_rate: 5 FPS (compressed), 100 FPS (reconstructed)

Applications & Benefits

• cell_imaging: Not specified
• benefits: Reduction of data bandwidth for video storage and transmission, real-time object detection and localization directly on compressed frames, enhancement in surveillance capabilities.

Supporting Organizations

• supported_by: Indian Institute of Science, tinyVision.ai

Manuscript Details

• publication_date: 19-22 Sept. 2021

Relevancy Score

• score: 9
• missing_fields:
  • dynamic_range
  • pixel_size
  • dark_current
  • focal_length
  • aperture
  • field_of_view
  • distortion
  • signal_to_noise_ratio
  • sensitivity
  • power_consumption
  • noise

Processed JSON Filename

request_13dc55f0-7d19-4b96-8d62-5f905520f23a-realtime_object_detection_and_localization_in_compressive_sensed_video.json