Deep learning methods have been applied in many domains with remarkable performance for the past decade, including in medical imaging. There is currently a growing interest in the research community for deep learning in Wireless Capsule Endoscopy (WCE). However, the availability of quality annotated dataset, and robust performing methods are still challenging in WCE. To fill this gap, MISAHUB provided an annotated dataset for bleeding in WCE, and launched the "AutoWCEBleedGen challenge Version V1" which aims at investigating deep learning methods for classification and detection of bleeding in WCE. As a response to the challenge, our work presents a method to classify bleeding and non-bleeding frames which is based on ResNet50 using cross validation and data augmentation techniques. Furthermore, we used YOLOv5 model to detect and draw the bounding boxes of the bleeding pixels in the bleeding images. Our solution achieves respectively 99.57% accuracy, 99.58% F1-score, and 99.53% Recall for the classification task on the validation set. For the detection task we achieve 70% AP, 68.20% mAP and 45% IoU. Our code is available at https://github.com/agossouema2011/ WCEBleedGenChallenge_Colorlab_Team.