In this video, we are going to see which is the best object detection algorithm or model for developers. We are going to test all the model based on three criterias: speed, accuracy and ease of implementation.

Object Detection is one of the most sought after domain of computer vision and the number of models available in this domain reflect the same, however not all models were create the same. Each model that we discuss in this video has its own pros and cons, but we are after those criterias that matters the most to us. We compared a model from the Two shot detector family which is Faster RCNN. The comparison also included two single shot models also, which are SSD (Single Shot Detectors) and YOLO. When comparing for speed, we focused on the inference speed of the models, ie how many frames can the model process in one second. For accuracy, we tried to see which model actually got the most accuracy and how reliable those accuracies are. Lastly, we also looked at the ease of implementation which basically focused on the framework (Opencv, PyTorch, TensorFlow) required to use the model and also the least number of lines of code we need to write to get the model to give detections.