search

nwesem / mtcnn_facenet_cpp_tensorRT

Face Recognition on NVIDIA Jetson (Nano) using TensorRT
GNU General Public License v3.0
200 stars 72 forks source link

Core at changing video stream from cam to rstp #17

Closed githubutilities closed 3 years ago

githubutilities commented 3 years ago

I just change cam to rstp videostream, and I got a core dump, which I trace it back to https://github.com/nwesem/mtcnn_facenet_cpp_tensorRT/blob/master/src/mtcnn.cpp#L85. Any idea how to fix it?

nwesem commented 3 years ago

hi @githubutilities, using a RSTP videostream as input would be a cool enhancement. Can you fork the repository and push your code such that I can take a look? Maybe I can make it work

githubutilities commented 3 years ago

It's simple. I just change line VideoStreamer videoStreamer = VideoStreamer("rtsp://192.168.10.21:8554/unicast", videoFrameWidth, videoFrameHeight); in file main.cpp. This can get frame from rstp video, but core at feeding the frame into facenet model.

jishang1602 commented 3 years ago

i tried with the VideoStreamer videoStreamer = VideoStreamer("rtsp://192.168.10.21:8554/unicast", videoFrameWidth, videoFrameHeight); in main.cpp but it is getting error which is shown below.

End generating TensorRT runtime models Parsing Directory: ../imgs Segmentation fault (core dumped)

in videoStreamer.h and videoStreamer.cpp VideoStreamer(int nmbrDevice, int videoWidth, int videoHeight, int frameRate, bool isCSICam); i changed with VideoStreamer(const char* nmbrDevice, int videoWidth, int videoHeight, int frameRate, bool isCSICam);

but getting the error Segmentation fault (core dumped)

nwesem commented 3 years ago

@jishang1602 Have you solved this? Maybe running it with gdb will help you find where it crashes. Currently, I don't have a setup to test this.

jishang1602 commented 3 years ago

@nwesem NO, it is still not solved. I have to convert the video to 640x480 size. May be this model is resized to 640x480. I have checked the https://github.com/PKUZHOU/MTCNN_FaceDetection_TensorRT/blob/dfad60565216a68413f434b500168c456fdd2587/src/demo.cpp file also. they are also doing resize of the frame. I need one more help from you. Can you help me in videoStreamer.cpp with threading? on rtsp the video is coming delayed. And after processing the delayed is increaded. My rtsp works on 20FPS. and the max FPS from the above project is 15-16 FPS. I need help on the real time processing. On Python i can do the threading and get the latest frame. In c++ i am unable to do that part. thank you

shubham-shahh commented 3 years ago

@nwesem NO, it is still not solved. I have to convert the video to 640x480 size. May be this model is resized to 640x480. I have checked the https://github.com/PKUZHOU/MTCNN_FaceDetection_TensorRT/blob/dfad60565216a68413f434b500168c456fdd2587/src/demo.cpp file also. they are also doing resize of the frame. I need one more help from you. Can you help me in videoStreamer.cpp with threading? on rtsp the video is coming delayed. And after processing the delayed is increaded. My rtsp works on 20FPS. and the max FPS from the above project is 15-16 FPS. I need help on the real time processing. On Python i can do the threading and get the latest frame. In c++ i am unable to do that part. thank you

how do you resize it?

jishang1602 commented 3 years ago

Once the frame is read from video or image.

E.g., frame = cv::imread (image_path); cv:: resize (frame,frame,cv:: Size (640,480),0,0,cv::INTER_CUBIC);

This way.

shubham-shahh commented 3 years ago

Once the frame is read from video or image.

E.g., frame = cv::imread (image_path); cv:: resize (frame,frame,cv:: Size (640,480),0,0,cv::INTER_CUBIC);

This way.

I am trying to take the input from rtsp stream