Human-Detection-from-Infrared-Images

Using Support Vector Machine and Extracting Histogram of Oriented Gradient Features. Future enhancements include:-

• Variable Bounding box
• Image Pyramid
• Model Implementation on Videos and Live feed
• More accuracy according to Available dataset and Computation Power.

Refer Tools and Resources for better understanding of Work Procedure

Dataset Sources

• "LSI-Far" Infrared Dataset (http://portal.uc3m.es/portal/page/portal/dpto_ing_sistemas_automatica/investigacion/IntelligentSystemsLab/research/InfraredDataset)

  • Increase Contrast in case, the images from LSI-Far will be taken in consideration.

• Infrared Dataset of Diverse Perspectives (http://vcipl-okstate.org/pbvs/bench/)

• "FLIR" Infrared Dataset (https://www.flir.in/oem/adas/adas-dataset-form/)

  • For the scenario of FLIR Dataset, Sign Up is free and then Dataset can be accessed (around 15.7 GB in total)

• "ASL" Infrared Dataset (https://projects.asl.ethz.ch/datasets/doku.php?id=ir:iricra2014)

• "Medeley" Infrared Image/Video Dataset (https://data.mendeley.com/datasets/n6wrv4ry6v/8)

References

• Dalal and Triggs Visible Band Imagery Research Paper (https://lear.inrialpes.fr/people/triggs/pubs/Dalal-cvpr05.pdf)
• Binary Bee Classifier (https://rpubs.com/Sharon_1684/454441)
• SciKit Documentation (https://scikit-learn.org/stable/modules/generated/sklearn.svm.SVC.html)
• Algorithms Information (https://www.pyimagesearch.com/)
• Debugging Help (https://stackoverflow.com/)
• Detailed Analysis of Various Classifiers on Infrared Image/Video Dataset:
  (https://www.researchgate.net/publication/343444436_Tuning_of_Classifiers_to_Speed-Up_Detection_of_Pedestrians_in_Infrared_Images)