From what I can see most object detection NNs (Fast(er) R-CNN, YOLO etc) are trained on data including bounding boxes indicating where in the picture the objects are localized.
Is there any model that can be trained with image+label data (without bounding box data) to predict the bounding box(es) ?
I am referring following papers
- Is object localization for free? – Weakly Supervised Object Recognition with Convolutional Neural Networks
- Self-taught object localization with deep networks
These papers are not recent and I am not able to find any recent papers related to this . Please let me know if there are any other resources related to available .
Thanks in advance