Low light object detection in outdoor scenes
Anush Kini
Unknown
- 0 Collaborators
The aim of this project is to detect objects like cars, two-wheelers, auto rickshaws, pedestrians, riders in low light conditions. An Exposure Fusion framework which enhances contrast of the image has been used for image enhancement. Currently, DeepLab-V3+ model is being used for segmentation. ...learn more
Project status: Concept
Intel Technologies
AI DevCloud / Xeon
Overview / Usage
The project aims to enhance the vision of the driver with smart algorithms that can work in low light scenarios – dusk, night, in lights from headlights etc. We hope to build a segmentation model that:
-
Can be used in real time situations.
-
Is affordable and can run in low cost hardware machines.
-
Detects objects on the road with a good accuracy.
Methodology / Approach
Our current software pipeline is listed below:
- Apply image enhancement algorithms before training
- Choose an appropriate deep learning model, implement and then train it on these enhanced images.
- During inference, apply image enhancement techniques.
- Inference then involves predicting the input data which might be in the form of an image or a video.
Image enhancement algorithms considered:
- Histogram equalisation
- Contrast Enhancement using Exposure Fusion framework
- Retinex based end to end deep learning model for image enhancement.
The Contrast Enhancement using Exposure Fusion framework provided the best mix of quality and speed between the three and hence, was adopted.