Open dumping of waste is a major global sustainability challenge and elimination of the practice is one of the main targets on the global agenda for sustainable development (UN, n.d., p. 6). In communities without systems for collection and disposal of waste materials, uncontrolled dumping remains the typical practice. It is estimated that three billion people worldwide lack access to controlled waste disposal facilities (Wilson et al., 2015), which presents serious consequences for natural ecosystems, human health, and economies. In Sub Saharan Africa for example, over 70 percent of the waste that is generated is openly disposed of in the environment (Ayeleru et al., 2020). On land such disposed waste materials are generally transported by rainwater to rivers, lakes and oceans, where they accumulate and harm natural ecosystems (Ostle et al., 2019; Zhu, 2021), specifically by causing death and physical damage to aquatic fauna through entanglement and ingestion (Gall & Thompson, 2015). Waste materials dumped in the environment also have the potential to present serious consequences for public health. Emerging studies indicate that waste materials such as plastics provide novel microhabitats for human pathogens (Gkoutselis et al., 2021; Rodrigues et al., 2019), and in 2022, a study showed for the first time the presence of microplastics in human blood (Leslie et al., 2022).
To curb open dumping of waste into the environment, several solutions have been suggested, such as the development and strict enforcement of legislation promoting household waste separation and collection, the development of adequate disposal facilities and implementation of waste recovery initiatives using a circular economy approach (Shi et al., 2021). Some countries have implemented a strict ban on the production and use of some products such as plastics (Nyathi & Togo, 2020; Xie & Martin, 2022), discouraging use of single use carrier bags, promoting waste clean-up campaigns, and introducing community waste recycling programs (Dlamini & Simatele, 2016). It is essential to assess the effectiveness of the implementation of these public health and environmental initiatives to reduce or eliminate uncontrolled dumping of waste.
Surveillance is a key approach to quantifying the problems associated with waste in the environment to enable policy makers to place it in context. Mapping of existing locations with disposed of waste is one approach to establishing where waste is being dumped and to understand whether waste mitigation strategies are working. This in turn will render the scale of this problem visible to policy makers. Mapping can be done using handheld Global Positioning Systems (GPS) to establish locations of waste piles. Mobile applications such as ‘Open Litter Maps’ allow users to capture geotagged photos which later enable mapping of locations where waste is being dumped (Lynch, 2018). However, the use of handheld GPS can only limit observations to locations that are physically accessible to the observer, and some dumpsites cannot be mapped. Additionally, it is difficult to quantify the spatial extent of existing waste piles. Aerial images on the other hand have the potential to overcome such limitations. For instance, satellite imagery has been used for the mapping of floating marine plastics at a global scale (Topouzelis et al., 2020). Still, most open satellite data has relatively coarse spatial resolution and it is difficult to use such data for mapping smaller waste piles, especially in the urban setting (Glanville & Chang, 2015). Even high-resolution optically satellite images, usually provided by private companies, are often affected by cloud cover and can be prohibitively expensive.
High-resolution aerial imagery captured by drones is a promising alternative. The use of drone imagery has been employed in previous studies, which have reported different approaches for mapping waste piles. One approach involves visual identification and manual labelling of objects present on the surface of the waste pile (Garcia-Garin et al., 2021; Jakovljevic et al., 2020; Pinto et al., 2021). Another approach involves manually identifying and labelling a small sample of waste piles or individual objects that are visible on the drone captured imagery and use these data as examples to train an image classification algorithm (Papakonstantinou et al., 2021; Wolf et al., 2020). The classification algorithms that have been previously employed include a segmentation threshold algorithm (Bao et al., 2018), random forest (Gonçalves, Andriolo, Gonçalves, et al., 2020; Gonçalves, Andriolo, Pinto, & Bessa, 2020; Gonçalves, Andriolo, Pinto, & Duarte, 2020; Martin et al., 2018), artificial neural networks (Pinto et al., 2021) and convolution neural networks (Fallati et al., 2019; Garcia-Garin et al., 2021; Gonçalves, Andriolo, Pinto, & Duarte, 2020; Jakovljevic et al., 2020; Kylili et al., 2019; Papakonstantinou et al., 2021; Wolf et al., 2020). These algorithms were applied on water surfaces and sandy beaches where there is a uniform background where it is relatively easy to discriminate and identify waste materials. In an urban environment with a non-uniform background, simple algorithms such as the segmentation threshold algorithm are unlikely to work well.
The aim of this study was to assess the practicalities of using drones for collection of high-resolution aerial imagery to map waste piles in an urban environment. We define waste pile as a collection of waste found in the environment, these might either been disposed of by humans or dispersed off by an agent such as stormwater or wind. We hypothesize that on aerial images, piles of waste formed by disposed of waste materials would exhibit distinct characteristics that might assist automatic mapping of waste piles from optical aerial images. We utilized the drone imagery to train classification algorithms to automate the detection of waste piles, and subsequently evaluated the performance of the detection workflow. To the best of our knowledge, this is the first application of low-cost drone imagery for mapping of waste piles along a river in Sub-Saharan Africa.