In the midst of the efforts in an item identification, region CNNs (rCNN) stands out as the most impressive, combining discriminatory exploration, CNNs, sustenance vector machines (SVM), and bounding box regression to achieve excellent object detection performance. We propose a new method for identifying numerous items from pictures using convolution neural nets (CNNs) in this presented study. The authors of the presented study use the edge box technique to create region suggestions from edge maps for each picture in our model, and then forward pass all of the proposals through a well-accepted CaffeNet prototype. Then we extract the yield of softmax that generally is most recent layer of CNN, to determine CNNs score for every proposal. One of the greedy suppression methodology referred to as non-maximum suppression (NMS) method is then used to combine the suggestions for each class separately. Finally, we assess each class's mean average precision (mAP). On the PASCAL 2007 test dataset, our model has a mAP of 37.38 percent. In this work, we also explore ways to enhance performance based on our model.