Object detection is used to detect the boxes on the path via the camera input.

          This feature requires continuous video stream from the camera. Code requires some libraries installed beforehand to execute. These libraries depend on the algorithm. If OpenCV is used with YOLOv5 ONNX format, OpenCV 4.5.4+ and GCC 9.0+ are required. If YOLOv5 is executed, then the requirements are written in its “requirements.txt” file.

            YOLOv5 needs to be retrained to find boxes. Therefore, there is a need for a dataset containing boxes with the proper annotation. It should be well-trained without any overfitting issues otherwise the algorithm produces unexpected results. After proper retraining, it can be used via OpenCV or its “detect.py” file.

        This feature is expected to detect boxes with high accuracy. There might be some limitations. The algorithm might struggle to detect the box because of external conditions such as angle and light. Since datasets are limited, the accuracy of the algorithm will be limited to the scope of its dataset.