Project Video Instructions
As we inch closer to the project submission timeline, we wanted to release a set of instructions to create a good presentation video. Please be mindful that these instructions only serve to be a guideline of what you aim to achieve. We do not mandate following the order or time breakdown perfectly, as long as you address all the high-level components. We have also linked the set of Top 25 video presentations from S'20 on the course website which can be found here.
Instructions:
Please try to adhere to the time limit, but also make sure you speak clearly and coherently so that your audience learns from your presentation. Total video length: 3 - 5 minutes (please try and not exceed the 5-minute limit, that said a few seconds here and there doesn't matter)
- The video can be slides with voice over but you can also get creative. There is no rule on who has to speak, it is left to your team's discretion.
- Use a screen - recording (screen-casting) tool. Most MacBooks and Windows laptops are equipped with an inbuilt one.
- The video must start with a slide of the title of the project, team-name, and names of group members.
- The video should be fluid enough so that you are able to convince an audience without any familiarity about your project topic, why your project topic is worth attempting, the goals you have accomplished, the methods deployed, the results achieved, and how were they similar/different to existing results, and why is it so!
Suggestive timeline of the video:
1. Description of the problem(20 - 50 seconds):
- Describe the problem at hand - What are you trying to solve, and why?
- Additionally, mention the significance of the problem - What do you think is the practical impact of your work?
- The novelty of the problem - Has the problem been solved before? If so, how are your results different from the current approach.
2. Task (<= 30 seconds):
- Mention whether your work is a re-implementation of a paper, application of an existing approach on a new dataset, or a novel approach.
- Talk about the dataset used, size of the datasets how some examples from the dataset in slides when possible in a tabulated fashion. A summary description of the dataset is very well appreciated. Also briefly mention any assumptions and any data transformations that were required.
3. Approach or Methods(1 minute - 2 minutes):
- Briefly describe the overall approach.
- Possibly include a flow-chart or diagram of your pipeline and/or architecture. Describe in detail the novel and/or the most significant components of your methods. Use diagrams wherever possible to make the presentation more comprehensible.
4. Results and Discussion (<= 1 minute):
- Describe your evaluation metrics and talk about the performance achieved.
- Report the best, interesting and unexpected results i.e no need to show the results for all your experiments. But don’t forget to emphasize on your takeaways after assessing these results - Why did you get those results? Why did it work better/worse than other approaches? Possibly include diagrams comparing your work and others' work. It is always good to answer the WHY for your results achieved and not just HOW?
5. Conclusions (<30 seconds):
- Briefly describe your conclusions from the observed results. Does your conclusion have a real-world impact?
A general note: please resist the urge to describe everything in great detail to avoid going over the time limit