The title of this question is explicit, I want to check the given map is injective, provided the field $\mathbb{K}$ is algebraically close; take for instance $\mathbb{K}=\mathbb{C}$ for clarification purposes.
My try: suppose $\phi(t_0:t_1)=\phi(s_0:s_1)$. Then, checking term by term, we obtain
$$t_0 = \lambda^{1/3}s_0^I$$ $$t_1=\lambda^{1/3}s_1^J$$
Where $I,J\in\{1,2,3\}$ are the superscrips of the distinct numbers in $\mathbb{K}$ for which the first and the third equation hold.
For simplicity, suppose $s_0=s_0^1$, $s_1 = s_1^1$. Substituting in the second equation:
$$t_0t_1^2=\lambda^{1/3}s_0^I \lambda^{2/3}{s_1^J}^2=\lambda s_0^I{s_1^J}^2=\lambda s_0^1{s_1^1}^2$$
In other words:
$$s_0^I{s_1^J}^2=s_0^1{s_1^1}^2$$
Which can only occur when $I=1$ and $J=1$, because taking square roots:
$$s_1^J=\pm s_1^1$$ And it would be impossible that $s_1^J=-s_1$, because then ${s_1^J}^3=-{s_1^1}^3$, contradicting the third equation, unless $s_1=0$, and in this case injectivity is clear.
Your proof is a bit confusing, so I'm not sure if it's correct. Here's the proof I would write:
Suppose $\phi(s_0:s_1) = \phi(t_0:t_1)$, and that $s_1 \neq 0$. It follows that (since we can divide by $s_1^2$) $$ (s_0s_1^2 :s_1^3) = (t_0t_1^2 :t_1^3) \implies\\ (s_0 : s_1) = (t_0t_1^2 :t_1^3) $$ Since $s_1 \neq 0$, conclude $t_1^3 \neq 0$, which means that $t_1 \neq 0$. Thus, we have $$ (s_0:s_1) = (t_0:t_1) $$ On the other hand, if $s_1 = 0$, then $\phi(s_0:s_1) = (s_0^3:0:0)$. If $\phi(t_0:t_1) = (s_0^3:0:0)$, then we have $t_0 = 0 \iff s_0 = 0$, which is to say that $(s_0:s_1) = (t_0:t_1)$ as desired.