What It Means to “See” Behaviors of Interest

What It Means to “See” Behaviors of Interest

Remember, the video camera is a filter. If the behaviors are not on camera, you cannot see them and you cannot code them; they might as well have never happened. Datavyu allows you to view your videos at various speeds. Exploit this feature to determine whether the behaviors of interest are adequately visible for coding.

On Camera

Camera views make all the difference. With well-designed camera views, the behaviors of interest are visible and large on the video frame. It can be advantageous to collect data from multiple camera views: Different behaviors may be visible from different angles, and it is often helpful to have multiple perspectives on a single behavior. Datavyu can synchronize views from different video files or you can use commercial software to merge and synchronize your camera views onto a single video frame. Often, the critical behavior of interest is obscured on one camera view; multiple camera views ensures that you will always be able to see the behavior.

Viewing Speed

Visibility varies depending on the viewing speed. For many behaviors, you will want to view the videos at speeds slower than real time. Datavyu provides easy, fingertip control over the viewing speed with the shuttle keys—allowing you to slow down, speed up, or pause the video with a tap of a finger. You can also view the videos frame by frame using fingertip control over the jog keys.

Rules of Thumb

Use HD video for a crisper, higher resolution image. But, be aware of data storage requirements. Most people don’t need the highest setting on their camera, which results in huge, unmanageable files. Large files also require more powerful computers for playback. The “lowest” HD setting is a lot easier to store and for most purposes looks great. Thus, consider the trade-off between higher resolution videos and storage/processing requirements. Your camera lenses and apertures will also affect whether the image is blurry, distorted, dark, or over-exposed.

In general, it is easier to code things you can see rather than things that you hear.

Use visual cues, not sound cues to demarcate sections of video. You can only hear sound while playing the video in real time and it is difficult to determine when a sound begins or ends with frame precision. Use visual contrast, not sound, to demarcate important sections of your recording session (e.g., new conditions, new trials). Abrupt changes in contrast are easier to see than subtle visual changes, especially at faster than real-time speeds (and you are likely to want to fast forward to different sections of your video). Flipping the room lights on and off provides an easily implemented and visible demarcation of sections of the session. Holding a bright card in front of a camera (with condition name or trial number) provides another easily implemented and visible demarcation of sections of the session.

Use multiple camera views to capture both small body parts/small movements and the whole body/scene. Small body parts, small movements, facial expressions, and eye movements are difficult to see when they are small in the video frame.

Part of planning your recording arrangement involves thinking about your recording context. For example, white skin is difficult to see against a light background and dark skin is difficult to see against a dark background. So, create more visual contrast by making your background a bright or saturated color or dressing the child in a bright or saturated color (we use bright blue carpets and mats because infants’ hands and feet are clearly visible against them).

Video Example

This video displays how easy it is to record videos that miss the behaviors of interest. The experimenter was interested in understanding how parents teach children to open containers. You can see by the video example that the single camera view is completely blocked by the parent.

Revise or add more camera views until you can thoroughly see your behaviors of interest.