my project about detecting the car and show is moving or not moving, so basically I have a CSV file that contains two labelings (moving and not moving) and I trained my model used a sequential model (fully connected layer) and save it with this Format hdf5. So my question now can I use my model (hdf5) with object detection algorithms like (Faster R-CNN, YOLO) if yes I hope can find a clear tutorial about that or my way is wrong and should I do with a different way
1 Answers
It's not clear how you want to use your model in conjunction with YOLO/Faster R-CNN, so I'll have to make some assumptions. It is also not clear from your question what is in your CSV file - is it actually pixel data or optical flow data? You possibly need to look up some easy YOLO tutorials first to see how it works.
YOLO is an object detection model. It takes an image as input and draws named bounding boxes around the objects it recognises in that image. It is fast enough to work real-time on some video. The input is pixels: an image (or video frame). The output is a series of labels and the co-ordinates of the bounding box (and most tutorials will draw the bounding boxes and labels on to the image and output that). You could use it to detect a car (using a pre-trained YOLO model file that identifies cars) and estimate co-ordinates to crop the video (spatially and temporally) so that you have just the right amount of information to run your model on. You would then need to get the pixels from your cropped video into the right format for your model (whatever that is). You can load a model from an hd5 model in keras using load_model("mymodelfilename.hd5")
Related