@MuradAlQurishee
Dear Mr. Jay, You are an awesome programmer. I did not buy books for learning deep learning. I saw all of your videos and waiting for new videos. My objects are different than the objects listed in the MASK RCNN. So I need to train the MASK RCNN in my local machine. I have a question below: Can you please make a video on Custom Mask RCNN model without using trained weight from the GitHub? I mean I have different objects than MASK RCNN used that needs train the MASK RCNN for my own case. It will be a matter of appreciation if you make the video from scratch that will suitable for different detecting any object training. Thanks
@ricardovelasqueznajera4111
Hey I have a question for you Mark; how would I connect an IP camera with this function. Also, thank you so much and congratulations for such amazing work Man; I've learned a lot from your videos.
@TheAllnun21
Thanks dude, you really saved my ass!!! Hope to learn more from you~
@paulchristian1244
Mark, once again awesome work! I have developed a program that uses one of several developed models (i.e. VGG-16, VGG-19, etc.) as the base model, and places a custom trained classifier (deep-NN) on top of the base model and fuse the two together to form a complete classification model for my specific objects. How can I do object detection and place the boxes on the object found? I started looking into “How to freeze a Graph” and the plan was to do this then run using the TF API. Have you experimented with this? Ideally, I’d like to just add new objects to current models, but that is not possible at this point – from what I can determine. This approach would at least enhance the usefullness of all the pre-defined models. Thanks, Mark!
@aiai-nn5jd
Thankyou mark! I solved the problem!
@magzhantinishbay536
hi mark , new to the field of machine learning and I would like to ask how you launched the atom in this video ?
@bestbyflyfm
can you demo on how to train a new weight to detect different thing?
@qizhao3539
Hello Mark, when I run the file visualize_cv.py, I got one problem: AssertionError: len(images) must be equal to BATCH_SIZE. Could you help me to figure out this problem. THX!
@Roytuhin
hey, Mark, I'm done with the real-time masking.. Thank you so much!! Now to extend further I want to remove the background of the masking image/video... CAN YOU HELP PLEASE? 🙏🙏🙏
@李枫-w8o
Thanks very much for your videos. They helped me a lot. But I have a problem when I run this code in the video:AttributeError: 'NoneType' object has no attribute 'shape'. I'm hoping you could do me a favor to sort it out.
@AhmedMohamed-wu8dp
Keep it up bro you make very useful videos
@PRASANNAMALITH-s1g
I want to apply a pure color to the identified objects instead of applying a mask. As an example, if a book has been identified by the Mask RCNN , then I want to add exactly 0,255,0 color to the objects pixels instead of a mask. Can you please show me a way to do it ?
@Augmented_AI
Hi Mark, any reason why you did not use the GPU for this demo? Or are you actually using GPU but the mask-RCNN is swamping the GPU?
@gohan8625
Hi! I want to detect only 1 specific class using the coco h5 file. How should I do it? For example I only want to detect person only without detecting the other classes
@abhinavsawant1862
Hey mark...which GPU u used for this?
@andrelanyak2131
ok done coding it, then how do you use it in jupyter, the video just cut to the test result with out showing what change in the jupyter notebook
@monjurulkarim6057
Hi, how can I set fixed color for individual class, so that, each object will have different color?
@ohhan777
Which IDE are you using? It looks great.
@TheSaggi007
Hey Mark I have got this error "ImportError: No module named 'coco'" . Can you suggest the solution?
@konstantinosvasili551