Problem Statement and solution
The main task is to "read every meter in a frame/image with accuracy". The whole task can be classified as a Computer vision task as all the data we will received in form of images or video.
The task can further be classified into 2 major tasks :
- Detection
- Segmentation
First and foremost we need to detect the meter in the frame as there might be one or more than one meters present in a single frame. Detection is one of the most basic computer vision task and there are a lot of approaches to tackle this, for this solution we have chosen Deep Learning because of the immense research and support available for object detection with DL.
In Deep Learning as well there are a lot of models and architecture catering the same purpose of object detection, it's always hard to pick one over the other. However, we have chosen EfficientDet-d0 for detection which is a scalable and efficient object detection model [1] and achieves state-of-the-art precision with fewer parameters.
Once we detected the meter in the frame and cropped it, the next step is to find scale and pointer and then use this information to calculate reading. For this we used the approach of semantic segmentation, which means we are going to classify every pixel in the image as one of the class which will lead to a segmentation map which can be used to locate pointer and scale hence calculating reading. Again for this there are many models like UNET, DeepLab, FCN etc. We chose UNET as the difference in inference time and accuracy is minimal[2].
Once we get the segmentation map last thing to do is to calculate reading with the provided information for which we first convert the circular meter segmentation to rectangle then calculate pointer location and scale location then relative location to calculate final readings.
- [1] EfficientDet: Scalable and Efficient Object Detection (2020)
- [2] Comparison of Deep-Learning-Based Segmentation Models: Using Top View Person Images (2020)
- [3] Automatic Meter Reading from UAV Inspection Photos in the Substation by Combining YOLOv5s and DeeplabV3+, 2022.