abhiTronix/vidgear
Vidgear
VidGear is a high-performance Python video processing framework designed for capturing, transcoding, and manipulating video streams. It functions as a multi-protocol video streamer and a WebRTC streaming server, enabling the transfer of video frames over networks using RTSP, RTMP, RTP, and MJPEG protocols.
The project distinguishes itself through hardware-accelerated video transcoding and decoding using GPU backends like CUDA to reduce CPU load. It includes a cross-platform screen capture tool and a specialized system for establishing direct peer-to-peer media connections using WebRTC signaling.
The framework covers a broad range of capabilities, including hardware video acquisition from USB cameras and Raspberry Pi modules, real-time video processing via filtergraph pipelines, and multi-protocol video routing. It also provides tools for adaptive streaming, network traffic encryption, and disk-based video recording.
The system includes utilities for the automatic installation of required FFmpeg binaries based on the operating system.
Features
- Real-Time Video Streaming Frameworks - Provides a comprehensive framework for capturing and broadcasting live video to web browsers and peer clients.
- Video Processing Frameworks - Implements a high-performance Python framework for capturing, transcoding, and manipulating video streams using FFmpeg.
- Cropped Screen Captures - Grabs frames from monitors in real-time using full-screen or specific area selection.
- Filtergraph Pipeline Management - Executes complex filtergraph pipelines to flip, scale, or modify video streams in real time during ingestion.
- Asynchronous Frame Queues - Processes video frames in a separate thread using a queue to prevent frame drops.
- Multi-Protocol Video Streamers - Provides a system for transferring video frames across multiple network protocols including RTSP, RTMP, RTP, and MJPEG.
- Real-Time Stream Transformations - Integrates custom processing logic to modify live video frames before transmitting them to remote peers.
- Filtergraph Pipelines - Processes raw video frames through linked filtergraph pipelines for real-time scaling, flipping, and stabilization.
- Hardware-Accelerated Video Pipelines - Utilizes hardware-accelerated video pipelines and complex filtergraphs for real-time media transformation.
- Network Video Ingestion - Reads live video frames from network URLs using standard protocols like HTTP, RTSP, and RTMP.
- Real-time Video Transcoding - Provides on-the-fly conversion of video formats and frames to ensure compatibility with receiving devices.
- Live Video Broadcasting - Streams high-quality adaptive video formats or MJPEG frames to web browsers and peer clients.
- Multi-Protocol Streaming - Pushes real-time frames to remote destinations using protocols like RTSP, RTP, and RTMP.
- Unified Source Abstractions - Provides a single interface to access various cameras and streams for easy source switching.
- Hardware-Accelerated Decoders - Offloads video decompression to CUDA or CUVID GPU backends to significantly reduce CPU utilization.
- Hardware-Specific Video Capture - Accesses hardware cameras using system indexes and platform-specific demuxers to retrieve raw video frames.
- Multi-Source Frame Ingestion - Interfaces with USB cameras, IP feeds, and screen captures for high-performance video frame ingestion.
- GPU-Accelerated Video Decoding - Offloads pixel processing and decompression to CUDA or CUVID backends to reduce CPU utilization.
- Hardware-Accelerated Transcoders - Leverages GPU hardware contexts like CUDA for high-performance video encoding and decoding.
- ICE-Based Peer Connection Managers - Establishes low-latency peer-to-peer media connections using ICE, STUN, and TURN signaling protocols.
- WebRTC Streaming - Provides a real-time asynchronous server for delivering audio and video streams to web browsers using WebRTC.
- Multi-Threaded Video Capture - Reads frames from IP cameras, network streams, and hardware decoders using multi-threaded processing.
- Non-Blocking Event Loops - Uses a non-blocking event-loop architecture to transfer high-speed video frames with minimal memory overhead.
- Screen Capture Tools - Includes a utility for grabbing real-time frames from monitors or specific screen areas for processing.
- Adaptive Segment Generation - Transcodes video files or real-time frames into chunked segments and manifest files for adaptive network playback.
- Disk Video Recording - Saves processed frames to a video file using backends to ensure valid output formatting.
- Image-to-Video Sequencing - Processes a series of images as a video stream using sequential patterns or globbing.
- Video Color Converters - Changes the color representation of a video source to meet specific processing or display requirements.
- Video File Encoding - Encodes real-time frames into compressed or uncompressed video files using specified bitrates and codecs.
- Adaptive Segmented Transcoding - Converts entire video files into a sequence of smaller segments to enable efficient delivery over HTTP.
- Subprocess-Based Pipelines - Executes heavy multimedia workloads, including decoding and filtergraphs, within isolated system subprocesses.
- Audio-Video Mixing - Combines real-time video frames with audio tracks into a single multimedia stream.
- Multi-Source Video Composition - Processes multiple simultaneous video inputs using complex filtergraphs to combine or manipulate streams.
- Software Video Stabilization - Removes jitter and perturbations from incoming video frames using a sliding-window of feature anchors.
- MJPEG Streaming - Streams live video frames to multiple web browser clients using a motion-JPEG delivery mechanism.
- Video Stream Extraction - Retrieves live video frames and associated metadata from third-party streaming platforms.
- Website Stream Pipelining - Pipelines live video frames and metadata from external streaming services through a specialized backend.
- URL-Based Video Pipelining - Streams video content directly from a URL address for processing without requiring manual downloads.
- Bidirectional Message Exchanges - Facilitates two-way communication streams for sending metadata and control signals between senders and receivers.
- High-Performance Data Transfer - Uses high-performance asynchronous messaging to stream video frames between systems with low latency.
- Asynchronous Event-Loop Streaming - Transfers high-speed video frames between servers and clients using asynchronous event loops to reduce system load.
- Chunked Segment Deliverers - Implements chunked segment delivery to enable adaptive network playback of real-time video streams.
- Traffic Distribution - Distributes video streams across network topologies using publish-subscribe or request-reply patterns.
- Camera and Video Capture - Interfaces with Raspberry Pi camera modules and USB webcams using multi-threaded wrappers.
- Audio and Video Processing - Provides a multi-threaded framework for video processing.
- Audio Video Processing - Listed in the “Audio Video Processing” section of the Awesome Python awesome list.
- Video Processing - Multi-threaded video processing framework.
