Kaamuk Shweta Cam Show Wid Facemp4 Install -

facemp4 \
  --device 0 \
  --output shweta_cam_if_face.mp4 \
  --detect-faces \
  --record-on-face \
  --min-face-duration 2   # start recording after a face has been present for 2 s

pip install --upgrade pip               # keep pip fresh
pip install facemp4[opencv]             # installs core + OpenCV for face detection

The [opencv] extra pulls in opencv-python-headless. If you want the GUI‑capable version (for local preview windows), install opencv-python instead:

pip uninstall opencv-python-headless
pip install opencv-python

# Inside the repo folder
mkdir -p build && cd build
# Configure – we enable the “wide‑angle” preset for the Kaamuk Shweta cam
cmake .. \
    -DCMAKE_BUILD_TYPE=Release \
    -DENABLE_WID_PRESET=ON \
    -DENABLE_RTMP=ON      # optional, for live streaming
# Compile (uses all CPU cores)
make -j$(nproc)
# Install system‑wide
sudo make install

After a successful install you’ll find the binaries in /usr/local/bin/:

If you enabled -DENABLE_RTMP=ON during the CMake step, you can push a live feed directly: kaamuk shweta cam show wid facemp4 install

facemp4-stream \
    -d /dev/video0 \
    -r 1920x1080 \
    -f 30 \
    -b 5M \
    -o rtmp://your.cdn/live/shweta

For RTSP (e.g., use with VLC, OBS, or a private Nginx‑rtsp server):

facemp4-stream \
    -d /dev/video0 \
    -r 1280x720 \
    -f 25 \
    -b 3M \
    -o rtsp://0.0.0.0:8554/shweta

You can add the above command to the daemon’s config (stream_url=) or launch it in a screen/tmux session. facemp4 \ --device 0 \ --output shweta_cam_if_face

=== QUICK INSTALL GUIDE ===
1️⃣ Prereqs
   - Windows 10/11 (64‑bit) / macOS 12+ / Ubuntu 20.04+
   - Install VC++ Redistributable (2015‑2022) & .NET 4.8 (Win)
   - Optional: FFmpeg (for extra codecs)
2️⃣ Download
   - Camera driver: https://kaamuk.com/shweta/drivers/kaamuk_shweta_win64.zip
   - FaceMP4:      https://facemp4.org/downloads/FaceMP4_v3.2.1_setup.exe
   - FFmpeg (optional): https://ffmpeg.org/download.html
3️⃣ Install driver
   - Extract → run setup.exe → reboot
4️⃣ Install FaceMP4
   - Run installer → default (or C:\Tools\FaceMP4) → allow firewall
5️⃣ Launch FaceMP4
   - Add Source → “USB Camera” → Device ID = 0
   - Set resolution 1280×720, FPS 30
   - Set output folder (e.g., C:\Videos\Cam1)
6️⃣ Test preview → you should see live video
7️⃣ (Optional) Enable continuous or motion‑triggered recording:
   Settings → Recording → Enable → set folder & naming pattern
8️⃣ Verify MP4: play with VLC, check smoothness
9️⃣ Troubleshoot
   - Black screen → wrong ID/driver
   - Crashes → install VC++ runtime
   - High CPU → enable HW accel, lower FPS

If you're looking to install software for viewing, editing, or processing video files (like MP4 files), there are several tools available depending on your operating system and specific needs. Here are some general suggestions:

| System | Main Language | Face Detector | License | Typical FPS (1080p) | |--------|----------------|----------------|---------|-------------------| | OpenCV‑VideoCapture + Haar | C++/Python | Haar cascade | BSD | 15–20 | | DeepStream SDK (NVIDIA) | C++ | SSD‑based DNN | Proprietary | 45+ | | Kaamuk Shweta CAM + FaceMP4 | C++/Qt | FaceMP4 (MobileNet‑SSD‑v2) | MIT | 30 | pip install --upgrade pip # keep pip fresh

FaceMP4 builds on the MobileNet‑SSD architecture but optimizes the inference path for MP4’s H.264/HEVC macro‑blocks, allowing detection directly on encoded frames. This technique, first described by Liu et al. (2022) in “Compressed‑Domain Object Detection”, reduces memory traffic by up to 40 %.

| Command | What it does | |---------|--------------| | facemp4 -h | Show full help/option list | | facemp4 --list-devices | Enumerate video (and audio) capture devices | | facemp4 --device 1 --output out.mp4 --duration 120 | Record 2 min from second webcam | | facemp4 --detect-faces --blur-faces | Live preview with blurred faces; MP4 saved unchanged | | python -m pip install facemp4[torch] | Install faster DNN‑based face detector | | ffmpeg -i out.mp4 -vf "crop=640:360:0:0" cropped.mp4 | Trim the video to a 640×360 region after recording |