Object detection performance is poor beyond 4m with SSD mobilenet using jetson inference

When running indoors, the SSD mobilenet detection model has poor performance past 4-6m. Outside is even worse. The objective is to find a way to improve the performance of object detection. Being able to detect objects up to 20m and beyond would add significant value. Some ways to do this may be to train or try a new model (yolo as an example), another way to do this may be to increase the resolution of the input images of the model (SSD trained on 300x300 images, retraining on 512x512 images should improve the performance for detecting objects further away).