CNN W3L05 : Bounding Box Predictions

CNN W3L05 : Bounding Box Predictions

Convolutional Neural Networks(CNN) Week 3 Lecture 5 : Bounding Box Predictions
**** Best Books on Machine Learning :
1. Introduction to Machine Learning with Python: A Guide for Data Scientists: https://amzn.to/2TLlhAR
2. Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow: Concepts, Tools, and Techniques to Build Intelligent Systems: https://amzn.to/2wKtPij
3. Pattern Recognition and Machine Learning (Information Science and Statistics): https://amzn.to/33aNXpL
4. Deep Learning with Python – François Chollet: https://amzn.to/39ISNgv
5. Deep Learning (Adaptive Computation and Machine Learning series) – Ian Goodfellow: https://amzn.to/2vMPVR7
6. Machine Learning: A Probabilistic Perspective (Adaptive Computation and Machine Learning series) – Kevin P. Murphy: https://amzn.to/33aNrYN
Watch the Reinforcement Learning course on Skillshare: https://skl.sh/2WHyoVG
Join Skillshare using this link to get 2 months free Premium Membership: https://skl.sh/2JQqXpi

#CNN #Bounding #Box

4 Antworten auf „CNN W3L05 : Bounding Box Predictions“

  1. From the description of the YOLO algorithm in this video, am I correct in thinking that the training data supplied should be such that each of these grid cells has sufficiently many objects appearing in them? For example, suppose the top-left grid cell does not have any objects appearing in it, in any of our training images. Then the network will not learn to adequately detect objects there, right?

    This leads me to think that the different grid cells don't quite share knowledge among one another. That is, they all have to be trained individually to some extent, by providing training data with objects within them.

    Thoughts, anyone?

  2. Why, in sliding window approach, matching exact position of an object is a problem? If the stride is 1, then we cover each pixel of the image (let's say with a 14×14 box centered at each pixel of the image), so we cover all the possible locations in image and therefore we will match the exact position of an object (its center). The problem arises only when we use a bigger stride.

Schreibe einen Kommentar

Deine E-Mail-Adresse wird nicht veröffentlicht. Erforderliche Felder sind mit * markiert.