Object detection based applications have become ubiquitous due to ever growing maturity of Deep learning technology and its democratization. There are however pitfalls with using readily available sources of data and algorithms for developing real world problem. In this whitepaper, the authors discuss the key challenges and their techniques to overcome them when developing an object detection application. The paper discusses in details various aspects related to data annotation, role of bias, performance vs accuracy trade-off, Explainable AI. Download to continue reading.
