A system for universal serial bus (USB) image capturing with two isochronous transfers includes a USB image capturing device and a host. The USB image capturing device has two endpoints for transferring a video stream through a USB. The host has a driver for receiving the video stream through the USB. The driver receives the video stream, and creates a read extension data structure for the endpoints. The driver manages and dispatches the packets of the endpoints, and copies data of the packets into a first buffer and a second buffer, respectively. When a first end frame flag in a packet of the first endpoint is set and a second end frame flag in a packet of the second endpoint is set, the driver combines data of the first and the second buffers into a frame and sends the frame to the host.